Datasets:

asoria

/

pdf-papers-docling

like 0

{ "children": [], "label": "unspecified", "name": "_root_", "self_ref": "#/furniture" }

{ "children": [ { "$ref": "#/texts/0" }, { "$ref": "#/texts/1" }, { "$ref": "#/texts/2" }, { "$ref": "#/texts/3" }, { "$ref": "#/texts/4" }, { "$ref": "#/texts/5" }, { "$ref": "#/texts/6" }, { "$ref": "#/texts/7" }, { "$ref": "#/texts/8" }, { "$ref": "#/texts/9" }, { "$ref": "#/texts/10" }, { "$ref": "#/texts/11" }, { "$ref": "#/texts/12" }, { "$ref": "#/texts/13" }, { "$ref": "#/texts/14" }, { "$ref": "#/texts/15" }, { "$ref": "#/tables/0" }, { "$ref": "#/texts/16" }, { "$ref": "#/pictures/0" }, { "$ref": "#/texts/17" }, { "$ref": "#/pictures/1" }, { "$ref": "#/texts/18" }, { "$ref": "#/texts/19" }, { "$ref": "#/texts/20" }, { "$ref": "#/texts/21" }, { "$ref": "#/texts/22" }, { "$ref": "#/texts/23" }, { "$ref": "#/texts/24" }, { "$ref": "#/texts/25" }, { "$ref": "#/texts/26" }, { "$ref": "#/texts/27" }, { "$ref": "#/texts/28" }, { "$ref": "#/texts/29" }, { "$ref": "#/texts/30" }, { "$ref": "#/texts/31" }, { "$ref": "#/texts/32" }, { "$ref": "#/texts/33" }, { "$ref": "#/texts/34" }, { "$ref": "#/texts/35" }, { "$ref": "#/texts/36" }, { "$ref": "#/texts/37" }, { "$ref": "#/texts/38" }, { "$ref": "#/texts/39" }, { "$ref": "#/texts/40" }, { "$ref": "#/texts/41" }, { "$ref": "#/texts/42" }, { "$ref": "#/texts/43" }, { "$ref": "#/texts/44" }, { "$ref": "#/texts/45" }, { "$ref": "#/texts/46" }, { "$ref": "#/texts/47" }, { "$ref": "#/texts/48" }, { "$ref": "#/texts/49" }, { "$ref": "#/texts/50" }, { "$ref": "#/texts/51" }, { "$ref": "#/texts/52" }, { "$ref": "#/texts/53" }, { "$ref": "#/texts/54" }, { "$ref": "#/texts/55" }, { "$ref": "#/texts/56" }, { "$ref": "#/pictures/2" }, { "$ref": "#/texts/57" }, { "$ref": "#/texts/58" }, { "$ref": "#/texts/59" }, { "$ref": "#/texts/60" }, { "$ref": "#/texts/61" }, { "$ref": "#/groups/0" }, { "$ref": "#/texts/68" }, { "$ref": "#/texts/69" }, { "$ref": "#/texts/70" }, { "$ref": "#/texts/71" }, { "$ref": "#/texts/72" }, { "$ref": "#/tables/1" }, { "$ref": "#/texts/73" }, { "$ref": "#/pictures/3" }, { "$ref": "#/texts/74" }, { "$ref": "#/texts/75" }, { "$ref": "#/texts/76" }, { "$ref": "#/texts/77" }, { "$ref": "#/texts/78" }, { "$ref": "#/texts/79" }, { "$ref": "#/texts/80" }, { "$ref": "#/texts/81" }, { "$ref": "#/texts/82" }, { "$ref": "#/texts/83" }, { "$ref": "#/texts/84" }, { "$ref": "#/pictures/4" }, { "$ref": "#/texts/85" }, { "$ref": "#/pictures/5" }, { "$ref": "#/texts/86" }, { "$ref": "#/texts/87" }, { "$ref": "#/texts/88" }, { "$ref": "#/texts/89" }, { "$ref": "#/texts/90" }, { "$ref": "#/texts/91" }, { "$ref": "#/texts/92" }, { "$ref": "#/texts/93" }, { "$ref": "#/texts/94" }, { "$ref": "#/pictures/6" }, { "$ref": "#/texts/95" }, { "$ref": "#/pictures/7" }, { "$ref": "#/texts/96" }, { "$ref": "#/texts/97" }, { "$ref": "#/texts/98" }, { "$ref": "#/texts/99" }, { "$ref": "#/texts/100" }, { "$ref": "#/texts/101" }, { "$ref": "#/texts/102" }, { "$ref": "#/pictures/8" }, { "$ref": "#/texts/103" }, { "$ref": "#/texts/104" }, { "$ref": "#/texts/105" }, { "$ref": "#/texts/106" }, { "$ref": "#/texts/107" }, { "$ref": "#/texts/108" }, { "$ref": "#/texts/109" }, { "$ref": "#/texts/110" }, { "$ref": "#/texts/111" }, { "$ref": "#/texts/112" }, { "$ref": "#/texts/113" }, { "$ref": "#/texts/114" }, { "$ref": "#/texts/115" }, { "$ref": "#/texts/116" }, { "$ref": "#/texts/117" }, { "$ref": "#/texts/118" }, { "$ref": "#/texts/119" }, { "$ref": "#/texts/120" }, { "$ref": "#/texts/121" }, { "$ref": "#/texts/122" }, { "$ref": "#/texts/123" }, { "$ref": "#/texts/124" }, { "$ref": "#/texts/125" }, { "$ref": "#/texts/126" }, { "$ref": "#/texts/127" }, { "$ref": "#/texts/128" }, { "$ref": "#/texts/129" }, { "$ref": "#/texts/130" }, { "$ref": "#/texts/131" }, { "$ref": "#/texts/132" }, { "$ref": "#/texts/133" }, { "$ref": "#/texts/134" }, { "$ref": "#/texts/135" }, { "$ref": "#/texts/136" }, { "$ref": "#/groups/1" }, { "$ref": "#/texts/138" }, { "$ref": "#/texts/139" }, { "$ref": "#/texts/140" }, { "$ref": "#/texts/141" }, { "$ref": "#/texts/142" }, { "$ref": "#/texts/143" }, { "$ref": "#/texts/144" }, { "$ref": "#/tables/2" }, { "$ref": "#/texts/145" }, { "$ref": "#/texts/146" }, { "$ref": "#/texts/147" }, { "$ref": "#/texts/148" }, { "$ref": "#/texts/149" }, { "$ref": "#/texts/150" }, { "$ref": "#/texts/151" }, { "$ref": "#/texts/152" }, { "$ref": "#/texts/153" }, { "$ref": "#/texts/154" }, { "$ref": "#/groups/2" }, { "$ref": "#/texts/156" }, { "$ref": "#/texts/157" }, { "$ref": "#/texts/158" }, { "$ref": "#/texts/159" }, { "$ref": "#/texts/160" }, { "$ref": "#/texts/161" }, { "$ref": "#/texts/162" }, { "$ref": "#/texts/163" }, { "$ref": "#/texts/164" }, { "$ref": "#/groups/3" }, { "$ref": "#/texts/166" }, { "$ref": "#/texts/167" }, { "$ref": "#/texts/168" }, { "$ref": "#/texts/169" }, { "$ref": "#/texts/170" }, { "$ref": "#/groups/4" }, { "$ref": "#/texts/173" }, { "$ref": "#/groups/5" }, { "$ref": "#/texts/175" }, { "$ref": "#/texts/176" }, { "$ref": "#/groups/6" }, { "$ref": "#/texts/178" }, { "$ref": "#/texts/179" }, { "$ref": "#/texts/180" }, { "$ref": "#/texts/181" }, { "$ref": "#/groups/7" }, { "$ref": "#/texts/183" }, { "$ref": "#/texts/184" }, { "$ref": "#/texts/185" }, { "$ref": "#/texts/186" }, { "$ref": "#/texts/187" }, { "$ref": "#/texts/188" }, { "$ref": "#/texts/189" }, { "$ref": "#/texts/190" }, { "$ref": "#/texts/191" }, { "$ref": "#/texts/192" }, { "$ref": "#/texts/193" }, { "$ref": "#/texts/194" }, { "$ref": "#/groups/8" }, { "$ref": "#/texts/196" }, { "$ref": "#/texts/197" }, { "$ref": "#/texts/198" }, { "$ref": "#/texts/199" }, { "$ref": "#/texts/200" }, { "$ref": "#/texts/201" }, { "$ref": "#/texts/202" }, { "$ref": "#/texts/203" }, { "$ref": "#/groups/9" }, { "$ref": "#/texts/205" }, { "$ref": "#/texts/206" }, { "$ref": "#/groups/10" }, { "$ref": "#/texts/209" }, { "$ref": "#/texts/210" }, { "$ref": "#/tables/3" }, { "$ref": "#/texts/211" }, { "$ref": "#/texts/212" }, { "$ref": "#/texts/213" }, { "$ref": "#/texts/214" }, { "$ref": "#/texts/215" }, { "$ref": "#/texts/216" }, { "$ref": "#/texts/217" }, { "$ref": "#/texts/218" }, { "$ref": "#/texts/219" }, { "$ref": "#/texts/220" }, { "$ref": "#/texts/221" }, { "$ref": "#/texts/222" }, { "$ref": "#/texts/223" }, { "$ref": "#/texts/224" }, { "$ref": "#/texts/225" }, { "$ref": "#/texts/226" }, { "$ref": "#/texts/227" }, { "$ref": "#/texts/228" }, { "$ref": "#/texts/229" }, { "$ref": "#/texts/230" }, { "$ref": "#/texts/231" }, { "$ref": "#/texts/232" }, { "$ref": "#/texts/233" }, { "$ref": "#/texts/234" }, { "$ref": "#/texts/235" }, { "$ref": "#/pictures/9" }, { "$ref": "#/texts/236" }, { "$ref": "#/texts/237" }, { "$ref": "#/texts/238" }, { "$ref": "#/texts/239" }, { "$ref": "#/texts/240" }, { "$ref": "#/texts/241" }, { "$ref": "#/texts/242" }, { "$ref": "#/texts/243" }, { "$ref": "#/texts/244" }, { "$ref": "#/texts/245" }, { "$ref": "#/groups/11" }, { "$ref": "#/texts/247" }, { "$ref": "#/texts/248" }, { "$ref": "#/groups/12" }, { "$ref": "#/texts/250" }, { "$ref": "#/texts/251" }, { "$ref": "#/texts/252" }, { "$ref": "#/texts/253" }, { "$ref": "#/texts/254" }, { "$ref": "#/texts/255" }, { "$ref": "#/texts/256" }, { "$ref": "#/texts/257" }, { "$ref": "#/texts/258" }, { "$ref": "#/tables/4" }, { "$ref": "#/texts/259" }, { "$ref": "#/texts/260" }, { "$ref": "#/texts/261" }, { "$ref": "#/pictures/10" }, { "$ref": "#/texts/262" }, { "$ref": "#/texts/263" }, { "$ref": "#/pictures/11" }, { "$ref": "#/texts/264" }, { "$ref": "#/pictures/12" }, { "$ref": "#/texts/265" }, { "$ref": "#/texts/266" }, { "$ref": "#/texts/267" }, { "$ref": "#/texts/268" }, { "$ref": "#/texts/269" }, { "$ref": "#/texts/270" }, { "$ref": "#/groups/13" }, { "$ref": "#/texts/288" }, { "$ref": "#/texts/289" }, { "$ref": "#/texts/290" }, { "$ref": "#/texts/291" }, { "$ref": "#/texts/292" }, { "$ref": "#/texts/293" }, { "$ref": "#/texts/294" }, { "$ref": "#/texts/295" }, { "$ref": "#/texts/296" }, { "$ref": "#/texts/297" }, { "$ref": "#/texts/298" }, { "$ref": "#/texts/299" }, { "$ref": "#/texts/300" }, { "$ref": "#/texts/301" }, { "$ref": "#/texts/302" }, { "$ref": "#/texts/303" }, { "$ref": "#/texts/304" }, { "$ref": "#/texts/305" }, { "$ref": "#/texts/306" }, { "$ref": "#/texts/307" }, { "$ref": "#/texts/308" }, { "$ref": "#/texts/309" }, { "$ref": "#/texts/310" }, { "$ref": "#/texts/311" }, { "$ref": "#/tables/5" }, { "$ref": "#/texts/312" }, { "$ref": "#/texts/313" }, { "$ref": "#/texts/314" }, { "$ref": "#/texts/315" }, { "$ref": "#/texts/316" }, { "$ref": "#/texts/317" }, { "$ref": "#/texts/318" }, { "$ref": "#/texts/319" }, { "$ref": "#/texts/320" }, { "$ref": "#/texts/321" }, { "$ref": "#/texts/322" }, { "$ref": "#/texts/323" }, { "$ref": "#/texts/324" }, { "$ref": "#/texts/325" }, { "$ref": "#/texts/326" }, { "$ref": "#/texts/327" }, { "$ref": "#/texts/328" }, { "$ref": "#/texts/329" }, { "$ref": "#/texts/330" }, { "$ref": "#/tables/6" }, { "$ref": "#/texts/331" }, { "$ref": "#/tables/7" }, { "$ref": "#/texts/332" }, { "$ref": "#/tables/8" }, { "$ref": "#/texts/333" }, { "$ref": "#/texts/334" }, { "$ref": "#/texts/335" }, { "$ref": "#/texts/336" }, { "$ref": "#/texts/337" }, { "$ref": "#/texts/338" }, { "$ref": "#/texts/339" }, { "$ref": "#/tables/9" }, { "$ref": "#/texts/340" }, { "$ref": "#/texts/341" }, { "$ref": "#/tables/10" }, { "$ref": "#/texts/342" }, { "$ref": "#/texts/343" }, { "$ref": "#/texts/344" }, { "$ref": "#/texts/345" }, { "$ref": "#/texts/346" }, { "$ref": "#/texts/347" }, { "$ref": "#/texts/348" }, { "$ref": "#/texts/349" }, { "$ref": "#/texts/350" }, { "$ref": "#/texts/351" }, { "$ref": "#/texts/352" }, { "$ref": "#/texts/353" }, { "$ref": "#/texts/354" }, { "$ref": "#/texts/355" }, { "$ref": "#/texts/356" }, { "$ref": "#/texts/357" }, { "$ref": "#/texts/358" }, { "$ref": "#/texts/359" }, { "$ref": "#/texts/360" }, { "$ref": "#/texts/361" }, { "$ref": "#/texts/362" }, { "$ref": "#/texts/363" }, { "$ref": "#/texts/364" }, { "$ref": "#/texts/365" }, { "$ref": "#/texts/366" }, { "$ref": "#/texts/367" }, { "$ref": "#/texts/368" }, { "$ref": "#/texts/369" }, { "$ref": "#/texts/370" }, { "$ref": "#/texts/371" }, { "$ref": "#/texts/372" }, { "$ref": "#/texts/373" }, { "$ref": "#/texts/374" }, { "$ref": "#/texts/375" }, { "$ref": "#/texts/376" }, { "$ref": "#/texts/377" }, { "$ref": "#/texts/378" }, { "$ref": "#/texts/379" }, { "$ref": "#/texts/380" }, { "$ref": "#/texts/381" }, { "$ref": "#/texts/382" }, { "$ref": "#/texts/383" }, { "$ref": "#/texts/384" }, { "$ref": "#/texts/385" }, { "$ref": "#/texts/386" }, { "$ref": "#/texts/387" }, { "$ref": "#/texts/388" }, { "$ref": "#/texts/389" }, { "$ref": "#/texts/390" }, { "$ref": "#/texts/391" }, { "$ref": "#/texts/392" }, { "$ref": "#/texts/393" }, { "$ref": "#/texts/394" }, { "$ref": "#/texts/395" }, { "$ref": "#/texts/396" }, { "$ref": "#/texts/397" }, { "$ref": "#/texts/398" }, { "$ref": "#/texts/399" }, { "$ref": "#/texts/400" }, { "$ref": "#/texts/401" } ], "label": "unspecified", "name": "_root_", "self_ref": "#/body" }

[ { "children": [ { "$ref": "#/texts/62" }, { "$ref": "#/texts/63" }, { "$ref": "#/texts/64" }, { "$ref": "#/texts/65" }, { "$ref": "#/texts/66" }, { "$ref": "#/texts/67" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/0" }, { "children": [ { "$ref": "#/texts/137" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/1" }, { "children": [ { "$ref": "#/texts/155" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/2" }, { "children": [ { "$ref": "#/texts/165" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/3" }, { "children": [ { "$ref": "#/texts/171" }, { "$ref": "#/texts/172" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/4" }, { "children": [ { "$ref": "#/texts/174" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/5" }, { "children": [ { "$ref": "#/texts/177" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/6" }, { "children": [ { "$ref": "#/texts/182" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/7" }, { "children": [ { "$ref": "#/texts/195" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/8" }, { "children": [ { "$ref": "#/texts/204" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/9" }, { "children": [ { "$ref": "#/texts/207" }, { "$ref": "#/texts/208" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/10" }, { "children": [ { "$ref": "#/texts/246" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/11" }, { "children": [ { "$ref": "#/texts/249" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/12" }, { "children": [ { "$ref": "#/texts/271" }, { "$ref": "#/texts/272" }, { "$ref": "#/texts/273" }, { "$ref": "#/texts/274" }, { "$ref": "#/texts/275" }, { "$ref": "#/texts/276" }, { "$ref": "#/texts/277" }, { "$ref": "#/texts/278" }, { "$ref": "#/texts/279" }, { "$ref": "#/texts/280" }, { "$ref": "#/texts/281" }, { "$ref": "#/texts/282" }, { "$ref": "#/texts/283" }, { "$ref": "#/texts/284" }, { "$ref": "#/texts/285" }, { "$ref": "#/texts/286" }, { "$ref": "#/texts/287" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/13" } ]

[ { "children": [], "enumerated": null, "label": "page_header", "level": null, "marker": null, "orig": "arXiv:2303.03915v1 [cs.CL] 7 Mar 2023", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 236.6788330078125, "coord_origin": "BOTTOMLEFT", "l": 17.525609970092773, "r": 36.33979415893555, "t": 576.9599609375 }, "charspan": [ 0, 37 ], "page_no": 1 } ], "self_ref": "#/texts/0", "text": "arXiv:2303.03915v1 [cs.CL] 7 Mar 2023" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "The BigScience ROOTS Corpus: A 1.6TB Composite Multilingual Dataset", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 654.7408447265625, "coord_origin": "BOTTOMLEFT", "l": 154.6439971923828, "r": 455.9135437011719, "t": 691.5006103515625 }, "charspan": [ 0, 67 ], "page_no": 1 } ], "self_ref": "#/texts/1", "text": "The BigScience ROOTS Corpus: A 1.6TB Composite Multilingual Dataset" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hugo Laurençon 1 ∗ Lucile Saulnier 1 ∗ Thomas Wang 1 ∗ Christopher Akiki 2 ∗ Albert Villanova del Moral 1 ∗ Teven Le Scao 1*", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 595.1638793945312, "coord_origin": "BOTTOMLEFT", "l": 131.032470703125, "r": 479.8136901855469, "t": 617.0003662109375 }, "charspan": [ 0, 124 ], "page_no": 1 } ], "self_ref": "#/texts/2", "text": "Hugo Laurençon 1 ∗ Lucile Saulnier 1 ∗ Thomas Wang 1 ∗ Christopher Akiki 2 ∗ Albert Villanova del Moral 1 ∗ Teven Le Scao 1*" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Leandro von Werra 1 Chenghao Mou 3 Eduardo González Ponferrada 4 Huu Nguyen 5 Jörg Frohberg 32 Mario Šaško 1 Quentin Lhoest 1", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 566.03759765625, "coord_origin": "BOTTOMLEFT", "l": 110.53775024414062, "r": 500.73431396484375, "t": 589.3616943359375 }, "charspan": [ 0, 125 ], "page_no": 1 } ], "self_ref": "#/texts/3", "text": "Leandro von Werra 1 Chenghao Mou 3 Eduardo González Ponferrada 4 Huu Nguyen 5 Jörg Frohberg 32 Mario Šaško 1 Quentin Lhoest 1" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Angelina McMillan-Major 1 , 6 Gérard Dupont 7 Stella Biderman 8 , 9 Anna Rogers 10 Loubna Ben allal 1 Francesco De Toni 11 Giada Pistilli 1 Olivier Nguyen 28 Somaieh Nikpoor 12 Maraim Masoud 13 Pierre Colombo 14 Javier de la Rosa 15 Paulo Villegas 16 Tristan Thrush 1 Shayne Longpre 17 Sebastian Nagel 19 Leon Weber 20 Manuel Romero Muñoz 21 Jian Zhu 22 Daniel van Strien 23 Zaid Alyafeai 24 Khalid Almubarak 25 Vu Minh Chien 26 Itziar Gonzalez-Dios 27 Aitor Soroa 27 Kyle Lo 29 Manan Dey 30 Pedro Ortiz Suarez 31 Aaron Gokaslan 18 Shamik Bose 3 David Ifeoluwa Adelani 33 Long Phan 34 Hieu Tran 34 Ian Yu 35 Suhas Pai 36 Jenny Chim 37", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 463.1165771484375, "coord_origin": "BOTTOMLEFT", "l": 108.63877868652344, "r": 503.3548278808594, "t": 561.831298828125 }, "charspan": [ 0, 634 ], "page_no": 1 } ], "self_ref": "#/texts/4", "text": "Angelina McMillan-Major 1 , 6 Gérard Dupont 7 Stella Biderman 8 , 9 Anna Rogers 10 Loubna Ben allal 1 Francesco De Toni 11 Giada Pistilli 1 Olivier Nguyen 28 Somaieh Nikpoor 12 Maraim Masoud 13 Pierre Colombo 14 Javier de la Rosa 15 Paulo Villegas 16 Tristan Thrush 1 Shayne Longpre 17 Sebastian Nagel 19 Leon Weber 20 Manuel Romero Muñoz 21 Jian Zhu 22 Daniel van Strien 23 Zaid Alyafeai 24 Khalid Almubarak 25 Vu Minh Chien 26 Itziar Gonzalez-Dios 27 Aitor Soroa 27 Kyle Lo 29 Manan Dey 30 Pedro Ortiz Suarez 31 Aaron Gokaslan 18 Shamik Bose 3 David Ifeoluwa Adelani 33 Long Phan 34 Hieu Tran 34 Ian Yu 35 Suhas Pai 36 Jenny Chim 37" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Violette Lepercq 1", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 446.7847900390625, "coord_origin": "BOTTOMLEFT", "l": 108.95610046386719, "r": 182.9969940185547, "t": 457.2112121582031 }, "charspan": [ 0, 18 ], "page_no": 1 } ], "self_ref": "#/texts/5", "text": "Violette Lepercq 1" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Suzana Ili'c 1", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 446.7847900390625, "coord_origin": "BOTTOMLEFT", "l": 195.9490966796875, "r": 246.76699829101562, "t": 457.2112121582031 }, "charspan": [ 0, 14 ], "page_no": 1 } ], "self_ref": "#/texts/6", "text": "Suzana Ili'c 1" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Margaret Mitchell 1", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 446.7847900390625, "coord_origin": "BOTTOMLEFT", "l": 259.7191162109375, "r": 341.77001953125, "t": 457.2112121582031 }, "charspan": [ 0, 19 ], "page_no": 1 } ], "self_ref": "#/texts/7", "text": "Margaret Mitchell 1" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Sasha Luccioni 1", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 446.7847900390625, "coord_origin": "BOTTOMLEFT", "l": 354.72210693359375, "r": 422.7060241699219, "t": 457.2112121582031 }, "charspan": [ 0, 16 ], "page_no": 1 } ], "self_ref": "#/texts/8", "text": "Sasha Luccioni 1" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Yacine Jernite 1", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 446.7847900390625, "coord_origin": "BOTTOMLEFT", "l": 435.6571044921875, "r": 499.697021484375, "t": 457.2112121582031 }, "charspan": [ 0, 16 ], "page_no": 1 } ], "self_ref": "#/texts/9", "text": "Yacine Jernite 1" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "$^{1}$Hugging Face $^{2}$Leipzig University $^{3}$Independent Researcher $^{4}$Ferrum Health $^{5}$Ontocord.ai $^{6}$University of Washington $^{7}$Mavenoid $^{8}$EleutherAI $^{9}$Booz Allen Hamilton $^{10}$University of Copenhagen $^{11}$University of Western Australia $^{12}$CAIDP $^{13}$Independent Researcher $^{14}$CentraleSupélec $^{15}$National Library of Norway $^{16}$Telefonica I+D $^{17}$MIT $^{18}$Cornell University $^{19}$Common Crawl $^{20}$Humboldt-Universität zu Berlin and Max Delbrück Center for Molecular Medicine $^{21}$Narrativa $^{22}$University of Michigan, Ann Arbor $^{23}$British Library $^{24}$King Fahd University of Petroleum and Minerals $^{25}$Prince Sattam bin Abdulaziz University (PSAU) $^{26}$DETOMO Inc. $^{27}$HiTZ Center, University of the Basque Country (UPV/EHU) $^{28}$ServiceNow $^{29}$Allen Institute for AI $^{30}$SAP $^{31}$Mannheim University $^{32}$Apergo.ai $^{33}$Saarland University $^{34}$VietAI Research $^{35}$Aggregate Intellect $^{36}$Bedrock AI $^{37}$Queen Mary University of London", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 310.32708740234375, "coord_origin": "BOTTOMLEFT", "l": 107.80108642578125, "r": 504.3467102050781, "t": 439.2336120605469 }, "charspan": [ 0, 1041 ], "page_no": 1 } ], "self_ref": "#/texts/10", "text": "$^{1}$Hugging Face $^{2}$Leipzig University $^{3}$Independent Researcher $^{4}$Ferrum Health $^{5}$Ontocord.ai $^{6}$University of Washington $^{7}$Mavenoid $^{8}$EleutherAI $^{9}$Booz Allen Hamilton $^{10}$University of Copenhagen $^{11}$University of Western Australia $^{12}$CAIDP $^{13}$Independent Researcher $^{14}$CentraleSupélec $^{15}$National Library of Norway $^{16}$Telefonica I+D $^{17}$MIT $^{18}$Cornell University $^{19}$Common Crawl $^{20}$Humboldt-Universität zu Berlin and Max Delbrück Center for Molecular Medicine $^{21}$Narrativa $^{22}$University of Michigan, Ann Arbor $^{23}$British Library $^{24}$King Fahd University of Petroleum and Minerals $^{25}$Prince Sattam bin Abdulaziz University (PSAU) $^{26}$DETOMO Inc. $^{27}$HiTZ Center, University of the Basque Country (UPV/EHU) $^{28}$ServiceNow $^{29}$Allen Institute for AI $^{30}$SAP $^{31}$Mannheim University $^{32}$Apergo.ai $^{33}$Saarland University $^{34}$VietAI Research $^{35}$Aggregate Intellect $^{36}$Bedrock AI $^{37}$Queen Mary University of London" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "∗ Equal contributions", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 293.3713073730469, "coord_origin": "BOTTOMLEFT", "l": 262.7587890625, "r": 348.76458740234375, "t": 304.93634033203125 }, "charspan": [ 0, 21 ], "page_no": 1 } ], "self_ref": "#/texts/11", "text": "∗ Equal contributions" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Abstract", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 266.7383728027344, "coord_origin": "BOTTOMLEFT", "l": 282.6378479003906, "r": 328.4288635253906, "t": 278.7103271484375 }, "charspan": [ 0, 8 ], "page_no": 1 } ], "self_ref": "#/texts/12", "text": "Abstract" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As language models grow ever larger, the need for large-scale high-quality text datasets has never been more pressing, especially in multilingual settings. The BigScience workshop, a 1-year international and multidisciplinary initiative, was formed with the goal of researching and training large language models as a values-driven undertaking, putting issues of ethics, harm, and governance in the foreground. This paper documents the data creation and curation efforts undertaken by BigScience to assemble the Responsible Open-science Open-collaboration Text Sources ( ROOTS ) corpus, a 1.6TB dataset spanning 59 languages that was used to train the 176-billion-parameter BigScience Large Open-science Open-access Multilingual ( BLOOM )(BigScience Workshop, 2022) language model. We further release a large initial subset of the corpus and analyses thereof, and hope to empower large-scale monolingual and multilingual modeling projects with both the data and the processing tools, as well as stimulate research around this large multilingual corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 95.36187744140625, "coord_origin": "BOTTOMLEFT", "l": 142.66099548339844, "r": 469.11334228515625, "t": 248.0401611328125 }, "charspan": [ 0, 1052 ], "page_no": 1 } ], "self_ref": "#/texts/13", "text": "As language models grow ever larger, the need for large-scale high-quality text datasets has never been more pressing, especially in multilingual settings. The BigScience workshop, a 1-year international and multidisciplinary initiative, was formed with the goal of researching and training large language models as a values-driven undertaking, putting issues of ethics, harm, and governance in the foreground. This paper documents the data creation and curation efforts undertaken by BigScience to assemble the Responsible Open-science Open-collaboration Text Sources ( ROOTS ) corpus, a 1.6TB dataset spanning 59 languages that was used to train the 176-billion-parameter BigScience Large Open-science Open-access Multilingual ( BLOOM )(BigScience Workshop, 2022) language model. We further release a large initial subset of the corpus and analyses thereof, and hope to empower large-scale monolingual and multilingual modeling projects with both the data and the processing tools, as well as stimulate research around this large multilingual corpus." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "36th Conference on Neural Information Processing Systems (NeurIPS 2022) Track on Datasets and Benchmarks.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 49.97235107421875, "coord_origin": "BOTTOMLEFT", "l": 107.57804870605469, "r": 505.75469970703125, "t": 58.769287109375 }, "charspan": [ 0, 105 ], "page_no": 1 } ], "self_ref": "#/texts/14", "text": "36th Conference on Neural Information Processing Systems (NeurIPS 2022) Track on Datasets and Benchmarks." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Contents", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.538330078125, "coord_origin": "BOTTOMLEFT", "l": 107.80149841308594, "r": 153.885498046875, "t": 718.8016357421875 }, "charspan": [ 0, 8 ], "page_no": 2 } ], "self_ref": "#/texts/15", "text": "Contents" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "2", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.96000289916992, "coord_origin": "BOTTOMLEFT", "l": 303.0603942871094, "r": 308.49029541015625, "t": 49.80267333984375 }, "charspan": [ 0, 1 ], "page_no": 2 } ], "self_ref": "#/texts/16", "text": "2" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 1: Overview of ROOTS. Left: A treemap of natural language representation in number of bytes by language family. The bulk of the graph is overwhelmed by the 1321.89 GB allotted to Eurasia. The orange rectangle corresponds to the 18GB of Indonesian, the sole representative of the Papunesia macroarea, and the green rectangle to the 0.4GB of the Africa linguistic macroarea. Right: A waffle plot of the distribution of programming languages by number of files. One square corresponds approximately to 30,000 files.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 495.5395202636719, "coord_origin": "BOTTOMLEFT", "l": 107.2073974609375, "r": 505.7450866699219, "t": 559.136474609375 }, "charspan": [ 0, 519 ], "page_no": 3 } ], "self_ref": "#/texts/17", "text": "Figure 1: Overview of ROOTS. Left: A treemap of natural language representation in number of bytes by language family. The bulk of the graph is overwhelmed by the 1321.89 GB allotted to Eurasia. The orange rectangle corresponds to the 18GB of Indonesian, the sole representative of the Papunesia macroarea, and the green rectangle to the 0.4GB of the Africa linguistic macroarea. Right: A waffle plot of the distribution of programming languages by number of files. One square corresponds approximately to 30,000 files." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "1 Introduction", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 461.96136474609375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 190.81365966796875, "t": 473.4407653808594 }, "charspan": [ 0, 14 ], "page_no": 3 } ], "self_ref": "#/texts/18", "text": "1 Introduction" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "BigScience 1 started in May 2021 as a one-year long open collaborative research initiative that gathered over a thousand participants around the world to study large language models (LLM). One of the founding goals of BigScience was to train an open-access, massively multilingual LLM, comparable in scale to GPT-3 (Brown et al., 2020) yet trained on a better documented and more representative multilingual dataset. The overall BigScience workshop was designed as a collaborative (Caselli et al., 2021; Bondi et al., 2021) and value-driven (Birhane et al., 2021) endeavor. Throughout the process of building this corpus we engaged in simultaneous investigation of ethical (Talat et al., 2022), sociopolitical (McMillan-Major et al., 2022), and data governance issues (Jernite et al., 2022) with the explicit goal of doing good for and by the people whose data we collected.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 351.38372802734375, "coord_origin": "BOTTOMLEFT", "l": 107.20756530761719, "r": 505.2413330078125, "t": 449.5952453613281 }, "charspan": [ 0, 874 ], "page_no": 3 } ], "self_ref": "#/texts/19", "text": "BigScience 1 started in May 2021 as a one-year long open collaborative research initiative that gathered over a thousand participants around the world to study large language models (LLM). One of the founding goals of BigScience was to train an open-access, massively multilingual LLM, comparable in scale to GPT-3 (Brown et al., 2020) yet trained on a better documented and more representative multilingual dataset. The overall BigScience workshop was designed as a collaborative (Caselli et al., 2021; Bondi et al., 2021) and value-driven (Birhane et al., 2021) endeavor. Throughout the process of building this corpus we engaged in simultaneous investigation of ethical (Talat et al., 2022), sociopolitical (McMillan-Major et al., 2022), and data governance issues (Jernite et al., 2022) with the explicit goal of doing good for and by the people whose data we collected." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Sourcing and building the dataset was organized around four working groups: Data Governance which helped define the project's values and design our approach to data usage and release in an international context, Data Sourcing and Preparation which was tasked with overseeing data collection, curation efforts, and Privacy for privacy risks and sanitizing the dataset, Legal Scholarship which helped define the multi-jurisdiction legal context in which the entire workshop was to operate, and we discuss practical implications throughout the paper where appropriate. An overview of the BigScience Corpus is provided in figure 1.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 269.4759521484375, "coord_origin": "BOTTOMLEFT", "l": 107.0902328491211, "r": 505.6561584472656, "t": 344.90972900390625 }, "charspan": [ 0, 627 ], "page_no": 3 } ], "self_ref": "#/texts/20", "text": "Sourcing and building the dataset was organized around four working groups: Data Governance which helped define the project's values and design our approach to data usage and release in an international context, Data Sourcing and Preparation which was tasked with overseeing data collection, curation efforts, and Privacy for privacy risks and sanitizing the dataset, Legal Scholarship which helped define the multi-jurisdiction legal context in which the entire workshop was to operate, and we discuss practical implications throughout the paper where appropriate. An overview of the BigScience Corpus is provided in figure 1." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The goal of the current paper is twofold: (1) we present a preliminary gated, subject to committing to the BigScience ethical charter$^{2}$, release of a large subset of ROOTS 3 (2) we release the numerous data tools 4 that were developed along the way and enabled us to curate, source, clean and inspect all 498 constituent datasets that come together to constitute ROOTS. This includes a preliminary results of the analyses that are currently being developed to study the corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 209.6044921875, "coord_origin": "BOTTOMLEFT", "l": 106.97081756591797, "r": 504.03564453125, "t": 263.3570556640625 }, "charspan": [ 0, 481 ], "page_no": 3 } ], "self_ref": "#/texts/21", "text": "The goal of the current paper is twofold: (1) we present a preliminary gated, subject to committing to the BigScience ethical charter$^{2}$, release of a large subset of ROOTS 3 (2) we release the numerous data tools 4 that were developed along the way and enabled us to curate, source, clean and inspect all 498 constituent datasets that come together to constitute ROOTS. This includes a preliminary results of the analyses that are currently being developed to study the corpus." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "1.1 Outline of the Paper", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 185.0235595703125, "coord_origin": "BOTTOMLEFT", "l": 107.9366683959961, "r": 216.99295043945312, "t": 194.96466064453125 }, "charspan": [ 0, 24 ], "page_no": 3 } ], "self_ref": "#/texts/22", "text": "1.1 Outline of the Paper" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The remainder of this paper details our approach to curating a web-scale dataset covering 59 languages, 46 natural languages and 13 programming languages - the language choice was chiefly driven by the communities who participated in the effort given the importance we placed on language expertise. Our final corpus is made up of two main components: 62% of the text comes from a community-selected and documented list of language data sources and its collection process is described in section 2, and", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 121.6297607421875, "coord_origin": "BOTTOMLEFT", "l": 107.03378295898438, "r": 505.2447814941406, "t": 174.810546875 }, "charspan": [ 0, 501 ], "page_no": 3 } ], "self_ref": "#/texts/23", "text": "The remainder of this paper details our approach to curating a web-scale dataset covering 59 languages, 46 natural languages and 13 programming languages - the language choice was chiefly driven by the communities who participated in the effort given the importance we placed on language expertise. Our final corpus is made up of two main components: 62% of the text comes from a community-selected and documented list of language data sources and its collection process is described in section 2, and" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{1}$https://bigscience.huggingface.co/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 101.934814453125, "coord_origin": "BOTTOMLEFT", "l": 122.1687240600586, "r": 249.4304656982422, "t": 111.841064453125 }, "charspan": [ 0, 40 ], "page_no": 3 } ], "self_ref": "#/texts/24", "text": "$^{1}$https://bigscience.huggingface.co/" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{2}$https://hf.co/spaces/bigscience/ethical-charter", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 91.7642593383789, "coord_origin": "BOTTOMLEFT", "l": 121.96000671386719, "r": 289.6260986328125, "t": 100.7054443359375 }, "charspan": [ 0, 53 ], "page_no": 3 } ], "self_ref": "#/texts/25", "text": "$^{2}$https://hf.co/spaces/bigscience/ethical-charter" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{3}$https://hf.co/bigscience-data", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 80.5430908203125, "coord_origin": "BOTTOMLEFT", "l": 122.12620544433594, "r": 227.3224639892578, "t": 90.22296142578125 }, "charspan": [ 0, 35 ], "page_no": 3 } ], "self_ref": "#/texts/26", "text": "$^{3}$https://hf.co/bigscience-data" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{4}$https://github.com/bigscience-workshop/data-preparation", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.85015869140625, "coord_origin": "BOTTOMLEFT", "l": 121.6886978149414, "r": 330.979736328125, "t": 79.129150390625 }, "charspan": [ 0, 61 ], "page_no": 3 } ], "self_ref": "#/texts/27", "text": "$^{4}$https://github.com/bigscience-workshop/data-preparation" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "3", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960086822509766, "coord_origin": "BOTTOMLEFT", "l": 302.8883972167969, "r": 308.49029541015625, "t": 49.4002685546875 }, "charspan": [ 0, 1 ], "page_no": 3 } ], "self_ref": "#/texts/28", "text": "3" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "38% consists of text extracted from a pre-processed web crawl, OSCAR (Ortiz Suárez et al. (2020)), filtered with the help of native speakers, which is described in section 3.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.8779296875, "coord_origin": "BOTTOMLEFT", "l": 107.63373565673828, "r": 505.2430725097656, "t": 717.632080078125 }, "charspan": [ 0, 174 ], "page_no": 4 } ], "self_ref": "#/texts/29", "text": "38% consists of text extracted from a pre-processed web crawl, OSCAR (Ortiz Suárez et al. (2020)), filtered with the help of native speakers, which is described in section 3." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "1.2 Related Work", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 671.5958251953125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 189.8626708984375, "t": 681.77978515625 }, "charspan": [ 0, 16 ], "page_no": 4 } ], "self_ref": "#/texts/30", "text": "1.2 Related Work" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Large Language Models and Large Text Corpora The current dominant paradigm in natural language processing relies heavily on pre-trained models: large language models that can then be fine-tuned on a downstream task (Howard and Ruder, 2018; Devlin et al., 2018) or even used as-is without additional data (Radford et al., 2019; Brown et al., 2020). In this paradigm, performance is directly correlated on both the model size and the dataset size and quality (Kaplan et al., 2020), with recent models trained on up to 1.4 trillion tokens (Hoffmann et al., 2022) and dataset creation pipelines representing a significant part of large language model projects. Most such datasets, however, are not released, hindering further research. Exceptions include the Pile (Gao et al., 2020), a curated corpus of datasets for language modeling that has become widely used for training state-of-the-art English-language models (Lieber et al., 2021; Smith et al., 2022; Black et al., 2022; Zhang et al., 2022), and C4 and mC4 (Raffel et al., 2020; Xue et al., 2020), which have powered the T5 family of models; CC100 (Conneau et al., 2020) which has seen heavy use for multilingual modeling; and OSCAR (Ortiz Suárez et al., 2019), which has enabled monolingual non-English models.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 520.2450561523438, "coord_origin": "BOTTOMLEFT", "l": 107.19105529785156, "r": 505.24713134765625, "t": 660.73388671875 }, "charspan": [ 0, 1265 ], "page_no": 4 } ], "self_ref": "#/texts/31", "text": "Large Language Models and Large Text Corpora The current dominant paradigm in natural language processing relies heavily on pre-trained models: large language models that can then be fine-tuned on a downstream task (Howard and Ruder, 2018; Devlin et al., 2018) or even used as-is without additional data (Radford et al., 2019; Brown et al., 2020). In this paradigm, performance is directly correlated on both the model size and the dataset size and quality (Kaplan et al., 2020), with recent models trained on up to 1.4 trillion tokens (Hoffmann et al., 2022) and dataset creation pipelines representing a significant part of large language model projects. Most such datasets, however, are not released, hindering further research. Exceptions include the Pile (Gao et al., 2020), a curated corpus of datasets for language modeling that has become widely used for training state-of-the-art English-language models (Lieber et al., 2021; Smith et al., 2022; Black et al., 2022; Zhang et al., 2022), and C4 and mC4 (Raffel et al., 2020; Xue et al., 2020), which have powered the T5 family of models; CC100 (Conneau et al., 2020) which has seen heavy use for multilingual modeling; and OSCAR (Ortiz Suárez et al., 2019), which has enabled monolingual non-English models." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Tooling, Visualization, and Replication Upstream from the finalized training datasets is the issue of processing methods and pipelines: both the operations that the datasets go through and the engineering effort required to apply them at terabyte scales. Existing work tends to fall on a spectrum from no details at all (Brown et al., 2020) to detailed filtering instructions, with (Raffel et al., 2020) or without the dataset release (Rae et al., 2021) to detailed filtering instructions with the accompanying code (Gao et al., 2020; Conneau et al., 2020; Ortiz Suárez et al., 2019). Even when the code is released, it tends to be built and tailored for the project's purpose. Consequently, large projects that do not re-use an existing dataset outright usually build their own pipeline rather than re-use an existing one on new data. However, data tools that were built and packaged in order to be used for other projects exist, such as OSCAR's Ungoliant and Goclassy (Abadji et al., 2021; Ortiz Suárez et al., 2019), which provides a distributed Common Crawl processing pipeline; CCNet (Wenzek et al., 2020), built for quality filtering of multilingual Common Crawl dumps; and OpenWebText (Gokaslan and Cohen, 2019), enabling Reddit dump processing.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 364.80377197265625, "coord_origin": "BOTTOMLEFT", "l": 107.08165740966797, "r": 505.2413330078125, "t": 505.4112854003906 }, "charspan": [ 0, 1252 ], "page_no": 4 } ], "self_ref": "#/texts/32", "text": "Tooling, Visualization, and Replication Upstream from the finalized training datasets is the issue of processing methods and pipelines: both the operations that the datasets go through and the engineering effort required to apply them at terabyte scales. Existing work tends to fall on a spectrum from no details at all (Brown et al., 2020) to detailed filtering instructions, with (Raffel et al., 2020) or without the dataset release (Rae et al., 2021) to detailed filtering instructions with the accompanying code (Gao et al., 2020; Conneau et al., 2020; Ortiz Suárez et al., 2019). Even when the code is released, it tends to be built and tailored for the project's purpose. Consequently, large projects that do not re-use an existing dataset outright usually build their own pipeline rather than re-use an existing one on new data. However, data tools that were built and packaged in order to be used for other projects exist, such as OSCAR's Ungoliant and Goclassy (Abadji et al., 2021; Ortiz Suárez et al., 2019), which provides a distributed Common Crawl processing pipeline; CCNet (Wenzek et al., 2020), built for quality filtering of multilingual Common Crawl dumps; and OpenWebText (Gokaslan and Cohen, 2019), enabling Reddit dump processing." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Documenting Textual Corpora in NLP An inspiration for our work is a recent emphasis on a more in-depth documentation of what is included and what is not in the corpora used for training NLP models . The most notable example of this is the Pile, for which the authors themselves analyze and document a variety of syntactic and semantic properties of the dataset including structural statistics (n-gram counts, language, document sizes), topical distributions across its components, social bias and sentiment co-occurrence, pejorative content, and information about licensing and authorial consent, in addition to releasing a datasheet (Biderman et al., 2022). Other LM pre-training datasets that have been documented and analyzed include C4 (Dodge et al., 2021; Luccioni and Viviano, 2021; Kreutzer et al., 2022), OSCAR (Kreutzer et al., 2022) and BookCorpus (Bandy and Vincent, 2021) . While this kind of documentation is far from standard practice, it is becoming increasingly common given recent calls for better documentation (Rogers, 2021; Bender et al., 2021) as well as empirical studies on data memorization in language models (Carlini et al., 2019, 2022).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 220.9249267578125, "coord_origin": "BOTTOMLEFT", "l": 107.10073852539062, "r": 505.2413024902344, "t": 350.8216552734375 }, "charspan": [ 0, 1163 ], "page_no": 4 } ], "self_ref": "#/texts/33", "text": "Documenting Textual Corpora in NLP An inspiration for our work is a recent emphasis on a more in-depth documentation of what is included and what is not in the corpora used for training NLP models . The most notable example of this is the Pile, for which the authors themselves analyze and document a variety of syntactic and semantic properties of the dataset including structural statistics (n-gram counts, language, document sizes), topical distributions across its components, social bias and sentiment co-occurrence, pejorative content, and information about licensing and authorial consent, in addition to releasing a datasheet (Biderman et al., 2022). Other LM pre-training datasets that have been documented and analyzed include C4 (Dodge et al., 2021; Luccioni and Viviano, 2021; Kreutzer et al., 2022), OSCAR (Kreutzer et al., 2022) and BookCorpus (Bandy and Vincent, 2021) . While this kind of documentation is far from standard practice, it is becoming increasingly common given recent calls for better documentation (Rogers, 2021; Bender et al., 2021) as well as empirical studies on data memorization in language models (Carlini et al., 2019, 2022)." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "2 (Crowd) Sourcing a Language Resource Catalogue", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 191.2431640625, "coord_origin": "BOTTOMLEFT", "l": 107.47827911376953, "r": 384.5262451171875, "t": 202.98883056640625 }, "charspan": [ 0, 48 ], "page_no": 4 } ], "self_ref": "#/texts/34", "text": "2 (Crowd) Sourcing a Language Resource Catalogue" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The first part of our corpus, accounting for 62% of the final dataset size (in bytes), was made up of a collection of monolingual and multilingual language resources that were selected and documented collaboratively through various efforts of the BigScience Data Sourcing working group. The first such effort consisted in creating a tool to support metadata collection through open submissions, called the BigScience Catalogue and running a series of hackathons in collaboration with locally-focused ML and NLP communities such as Masakhane, Machine Learning Tokyo and LatinX in AI where participants could add and document entries for their languages to the catalogue (McMillan-Major et al., 2022). This yielded a set of 252 sources, including at least 21 per considered language category. We focused on metadata collection as a way to support selection of the sources for the final dataset and documentation of the final dataset. In parallel, working group participants gathered additional", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.72589111328125, "coord_origin": "BOTTOMLEFT", "l": 107.1939926147461, "r": 505.73931884765625, "t": 177.3785400390625 }, "charspan": [ 0, 991 ], "page_no": 4 } ], "self_ref": "#/texts/35", "text": "The first part of our corpus, accounting for 62% of the final dataset size (in bytes), was made up of a collection of monolingual and multilingual language resources that were selected and documented collaboratively through various efforts of the BigScience Data Sourcing working group. The first such effort consisted in creating a tool to support metadata collection through open submissions, called the BigScience Catalogue and running a series of hackathons in collaboration with locally-focused ML and NLP communities such as Masakhane, Machine Learning Tokyo and LatinX in AI where participants could add and document entries for their languages to the catalogue (McMillan-Major et al., 2022). This yielded a set of 252 sources, including at least 21 per considered language category. We focused on metadata collection as a way to support selection of the sources for the final dataset and documentation of the final dataset. In parallel, working group participants gathered additional" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "4", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.44378662109375, "r": 308.49029541015625, "t": 49.5997314453125 }, "charspan": [ 0, 1 ], "page_no": 4 } ], "self_ref": "#/texts/36", "text": "4" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Arabic language resources in the Masader repository (Alyafeai et al., 2021), and proposed a list of websites of interest to increase the geographical diversity of our English, Spanish, and Chinese language data. Finally, in order to explicitly test large language models' ability to handle computer code along with natural language, we selected code data available on GitHub and StackExchange.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 674.6559448242188, "coord_origin": "BOTTOMLEFT", "l": 107.08114624023438, "r": 504.1871032714844, "t": 717.5919189453125 }, "charspan": [ 0, 393 ], "page_no": 5 } ], "self_ref": "#/texts/37", "text": "Arabic language resources in the Masader repository (Alyafeai et al., 2021), and proposed a list of websites of interest to increase the geographical diversity of our English, Spanish, and Chinese language data. Finally, in order to explicitly test large language models' ability to handle computer code along with natural language, we selected code data available on GitHub and StackExchange." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "2.1 Obtaining Data from the Identified Resources", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 648.3638916015625, "coord_origin": "BOTTOMLEFT", "l": 107.70671081542969, "r": 324.75067138671875, "t": 658.2896728515625 }, "charspan": [ 0, 48 ], "page_no": 5 } ], "self_ref": "#/texts/38", "text": "2.1 Obtaining Data from the Identified Resources" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Gathering Identified Datasets and Collections. First, we leveraged the BigScience Catalogue and the Masader repository to start obtaining text from identified sources, which included both existing NLP datasets and collections of documents of various compositions. Given the diversity of sources, hosting methods, data custodians, and formats, collecting this text required a collaborative effort. To that end, we established a 2-phase approach: first, collect as many data sources as possible in an easily accessible location; second, map all of them to a common format to ease further processing.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 572.7410888671875, "coord_origin": "BOTTOMLEFT", "l": 107.029296875, "r": 505.7449035644531, "t": 637.640625 }, "charspan": [ 0, 597 ], "page_no": 5 } ], "self_ref": "#/texts/39", "text": "Gathering Identified Datasets and Collections. First, we leveraged the BigScience Catalogue and the Masader repository to start obtaining text from identified sources, which included both existing NLP datasets and collections of documents of various compositions. Given the diversity of sources, hosting methods, data custodians, and formats, collecting this text required a collaborative effort. To that end, we established a 2-phase approach: first, collect as many data sources as possible in an easily accessible location; second, map all of them to a common format to ease further processing." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In the first phase, we organized an open hackathon to start gathering identified sources on the Hugging Face Datasets hub (Lhoest et al., 2021), in a dedicated organization 5 (in order to manage access controls). In the second phase, the collected datasets were furthered processed via (1) Language segmentation , whereby data sources were split using metadata for each covered language in order to obtain monolingual datasets, and the use of (2) Uniform interface whereby a document consisted of two fields: \"text\" for the actual text content, and \"meta\" with a JSON representation of metadata for a given document, containing sufficient information to trace documents back to their original sources.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 491.4092712402344, "coord_origin": "BOTTOMLEFT", "l": 107.0983657836914, "r": 505.74688720703125, "t": 566.7156982421875 }, "charspan": [ 0, 701 ], "page_no": 5 } ], "self_ref": "#/texts/40", "text": "In the first phase, we organized an open hackathon to start gathering identified sources on the Hugging Face Datasets hub (Lhoest et al., 2021), in a dedicated organization 5 (in order to manage access controls). In the second phase, the collected datasets were furthered processed via (1) Language segmentation , whereby data sources were split using metadata for each covered language in order to obtain monolingual datasets, and the use of (2) Uniform interface whereby a document consisted of two fields: \"text\" for the actual text content, and \"meta\" with a JSON representation of metadata for a given document, containing sufficient information to trace documents back to their original sources." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Pseudo-Crawled Data. Of the various categories of language resources identified through the data sourcing effort, websites stood out as one that required a particular effort and dedicated pipeline. We decided to design such a pipeline based on \"pseudo-crawling\": that is, rather than crawling the websites ourselves, we retrieved pages corresponding to the target domain names from 18 snapshots archived by Common Crawl in 2020 and 2021 in Web ARChive (WARC) format (Mohr et al., 2008). These domain names came from two main sources: the homepage field in the metadata of the 252 above-mentioned catalogue entries when available (192 in total), and the 456 websites proposed by participants asynchronously to improve the geographical diversity of our language sources; which yielded a total of 614 unique domain names after deduplication.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 378.81561279296875, "coord_origin": "BOTTOMLEFT", "l": 106.80099487304688, "r": 505.74493408203125, "t": 476.1231384277344 }, "charspan": [ 0, 838 ], "page_no": 5 } ], "self_ref": "#/texts/41", "text": "Pseudo-Crawled Data. Of the various categories of language resources identified through the data sourcing effort, websites stood out as one that required a particular effort and dedicated pipeline. We decided to design such a pipeline based on \"pseudo-crawling\": that is, rather than crawling the websites ourselves, we retrieved pages corresponding to the target domain names from 18 snapshots archived by Common Crawl in 2020 and 2021 in Web ARChive (WARC) format (Mohr et al., 2008). These domain names came from two main sources: the homepage field in the metadata of the 252 above-mentioned catalogue entries when available (192 in total), and the 456 websites proposed by participants asynchronously to improve the geographical diversity of our language sources; which yielded a total of 614 unique domain names after deduplication." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We collected URLs contained within those domains using the Common Crawl index. The index provides metadata for every document including the page URL, WARC filename and record offsets, fetch status, content MIME type, etc. We ran a query matching all documents that share the domain name with a seed using Amazon Athena on Common Crawl's columnar index$^{6}$. 48 of the 614 initial seed domain names had no matches in the index and were therefore left out. Once we obtained the document metadata, we fetched the WARC records using HTTP range requests with the start and end byte offsets. Since HTML web pages constitute the largest portion of pages contained in the Common Crawl dumps, we decided to only extract text from HTML pages. Documents in other formats were filtered out, ie XML, PDF, etc. 27 domain names were additionally removed from the list at this stage as we had not retrieved any HTML pages for them.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 263.8909912109375, "coord_origin": "BOTTOMLEFT", "l": 107.11609649658203, "r": 505.23968505859375, "t": 372.3113098144531 }, "charspan": [ 0, 916 ], "page_no": 5 } ], "self_ref": "#/texts/42", "text": "We collected URLs contained within those domains using the Common Crawl index. The index provides metadata for every document including the page URL, WARC filename and record offsets, fetch status, content MIME type, etc. We ran a query matching all documents that share the domain name with a seed using Amazon Athena on Common Crawl's columnar index$^{6}$. 48 of the 614 initial seed domain names had no matches in the index and were therefore left out. Once we obtained the document metadata, we fetched the WARC records using HTTP range requests with the start and end byte offsets. Since HTML web pages constitute the largest portion of pages contained in the Common Crawl dumps, we decided to only extract text from HTML pages. Documents in other formats were filtered out, ie XML, PDF, etc. 27 domain names were additionally removed from the list at this stage as we had not retrieved any HTML pages for them." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "To extract the text from the HTML pages, we first minified the HTML code. Minification is the removal of unnecessary characters from the source code of a website. Inspired by Aghajanyan et al. (2022), we removed from the DOM-HTML all the sub-trees contained in a <script> , <style> , <header> , <iframe> , <footer> and <form> tag as well as all the sub-trees associated with a <body> , <div> , <p> , <section> , <table> , <ul> , <ol> or <dl> tag whose textual content was less than 64 characters long. The text was then extracted from the nodes of this new DOM-HTML. While concatenating the text extracted, we applied a set of rules to reconstruct the structure of the text without its HTML code, inspired by what Common Crawl does to extract its WET files (Appendix B.1). The overall procedure enabled us to obtain text datasets for 539 domain names.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 160.89108276367188, "coord_origin": "BOTTOMLEFT", "l": 105.97799682617188, "r": 505.2451477050781, "t": 257.56719970703125 }, "charspan": [ 0, 851 ], "page_no": 5 } ], "self_ref": "#/texts/43", "text": "To extract the text from the HTML pages, we first minified the HTML code. Minification is the removal of unnecessary characters from the source code of a website. Inspired by Aghajanyan et al. (2022), we removed from the DOM-HTML all the sub-trees contained in a <script> , <style> , <header> , <iframe> , <footer> and <form> tag as well as all the sub-trees associated with a <body> , <div> , <p> , <section> , <table> , <ul> , <ol> or <dl> tag whose textual content was less than 64 characters long. The text was then extracted from the nodes of this new DOM-HTML. While concatenating the text extracted, we applied a set of rules to reconstruct the structure of the text without its HTML code, inspired by what Common Crawl does to extract its WET files (Appendix B.1). The overall procedure enabled us to obtain text datasets for 539 domain names." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "GitHub Code. We collected a code dataset from BigQuery 7 using the same language selection as AlphaCode (Li et al., 2022). The dataset was then deduplicated of exact matches and filtered for source files with between 100 and 200,000 characters, between 15-65% alphabetic characters, a max", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 113.94806671142578, "coord_origin": "BOTTOMLEFT", "l": 107.05652618408203, "r": 504.3546447753906, "t": 146.2622528076172 }, "charspan": [ 0, 288 ], "page_no": 5 } ], "self_ref": "#/texts/44", "text": "GitHub Code. We collected a code dataset from BigQuery 7 using the same language selection as AlphaCode (Li et al., 2022). The dataset was then deduplicated of exact matches and filtered for source files with between 100 and 200,000 characters, between 15-65% alphabetic characters, a max" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{5}$https://hf.co/bigscience-catalogue-data", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 90.855224609375, "coord_origin": "BOTTOMLEFT", "l": 121.87751770019531, "r": 264.6342468261719, "t": 101.43792724609375 }, "charspan": [ 0, 45 ], "page_no": 5 } ], "self_ref": "#/texts/45", "text": "$^{5}$https://hf.co/bigscience-catalogue-data" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{6}$https://commoncrawl.org/2018/03/index-to-warc-files-and-urls-in-columnar-format/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 80.75225830078125, "coord_origin": "BOTTOMLEFT", "l": 121.34783935546875, "r": 427.1984558105469, "t": 89.46319580078125 }, "charspan": [ 0, 86 ], "page_no": 5 } ], "self_ref": "#/texts/46", "text": "$^{6}$https://commoncrawl.org/2018/03/index-to-warc-files-and-urls-in-columnar-format/" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{7}$“GitHub on BigQuery: Analyze all the open source code”", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 70.06011962890625, "coord_origin": "BOTTOMLEFT", "l": 122.33873748779297, "r": 334.24493408203125, "t": 80.12591552734375 }, "charspan": [ 0, 60 ], "page_no": 5 } ], "self_ref": "#/texts/47", "text": "$^{7}$“GitHub on BigQuery: Analyze all the open source code”" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "5", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960086822509766, "coord_origin": "BOTTOMLEFT", "l": 302.7445068359375, "r": 308.49029541015625, "t": 49.4913330078125 }, "charspan": [ 0, 1 ], "page_no": 5 } ], "self_ref": "#/texts/48", "text": "5" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "line length of 20-1000 characters, and a token length standard deviation of more than 3. Due to a bug in the pre-processing pipeline the dataset was also filtered for GPL licenses only.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.8427734375, "coord_origin": "BOTTOMLEFT", "l": 107.54682922363281, "r": 503.9951171875, "t": 717.70849609375 }, "charspan": [ 0, 185 ], "page_no": 6 } ], "self_ref": "#/texts/49", "text": "line length of 20-1000 characters, and a token length standard deviation of more than 3. Due to a bug in the pre-processing pipeline the dataset was also filtered for GPL licenses only." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Merging and Deduplicating Sources. After gathering and processing language data via the three pipelines outlined above, we took a final step to manually inspect, deduplicate, and make a further selection of the sources. First, we addressed dataset overlap we found by looking through our sources. For example: OpenITI was present in both its raw form as well as a processed version. Consensus was reached to choose the latter version. Non-trivial datasets overlap included s2orc (Lo et al., 2020), Arxiv (Clement et al., 2019) and the PubMed Central subset of the Pile (Gao et al., 2020). We also performed cross-pipeline dataset deduplication, removing the pseudo-crawled Wikipedia and GitHub in favor of their other versions. We also removed datasets that we found had a high incidence of documents that were not fully in natural language (e.g. unexpected instances of SEO, HTML tags etc...), as well as very small datasets in the higher-resourced languages. Finally, pseudo-crawled sources were further processed to remove menus (with a heuristic consisting of removing lines that occurred in more than 1% of pages in a given domain) and pages that had a high incidence of character ngram repetition, low language identification confidence, or low proportion of closed class words (see Section 3). We then removed entire domains whose size was less than 2MB after this step, yielding 147 pseudo-crawl-based datasets, and a total of 517 datasets including all three pipelines.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 492.2505798339844, "coord_origin": "BOTTOMLEFT", "l": 106.89788055419922, "r": 505.73931884765625, "t": 655.0146484375 }, "charspan": [ 0, 1478 ], "page_no": 6 } ], "self_ref": "#/texts/50", "text": "Merging and Deduplicating Sources. After gathering and processing language data via the three pipelines outlined above, we took a final step to manually inspect, deduplicate, and make a further selection of the sources. First, we addressed dataset overlap we found by looking through our sources. For example: OpenITI was present in both its raw form as well as a processed version. Consensus was reached to choose the latter version. Non-trivial datasets overlap included s2orc (Lo et al., 2020), Arxiv (Clement et al., 2019) and the PubMed Central subset of the Pile (Gao et al., 2020). We also performed cross-pipeline dataset deduplication, removing the pseudo-crawled Wikipedia and GitHub in favor of their other versions. We also removed datasets that we found had a high incidence of documents that were not fully in natural language (e.g. unexpected instances of SEO, HTML tags etc...), as well as very small datasets in the higher-resourced languages. Finally, pseudo-crawled sources were further processed to remove menus (with a heuristic consisting of removing lines that occurred in more than 1% of pages in a given domain) and pages that had a high incidence of character ngram repetition, low language identification confidence, or low proportion of closed class words (see Section 3). We then removed entire domains whose size was less than 2MB after this step, yielding 147 pseudo-crawl-based datasets, and a total of 517 datasets including all three pipelines." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "2.2 Processing Pipeline for Quality Improvement on Crowdsourced Datasets", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 438.81317138671875, "coord_origin": "BOTTOMLEFT", "l": 107.43009948730469, "r": 438.240234375, "t": 448.8727111816406 }, "charspan": [ 0, 72 ], "page_no": 6 } ], "self_ref": "#/texts/51", "text": "2.2 Processing Pipeline for Quality Improvement on Crowdsourced Datasets" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Once a text field was obtained, we attempted to improve the quality of that text. In the specific case of text extraction from HTML, we observe that not all text are relevant (menus, advertisements, repeated text on each page etc ...). In order to remove noisy data from our dataset, we applied a processing pipeline for each dataset consisting of a sequence of functions.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 375.0260314941406, "coord_origin": "BOTTOMLEFT", "l": 107.39774322509766, "r": 504.33404541015625, "t": 417.5008544921875 }, "charspan": [ 0, 372 ], "page_no": 6 } ], "self_ref": "#/texts/52", "text": "Once a text field was obtained, we attempted to improve the quality of that text. In the specific case of text extraction from HTML, we observe that not all text are relevant (menus, advertisements, repeated text on each page etc ...). In order to remove noisy data from our dataset, we applied a processing pipeline for each dataset consisting of a sequence of functions." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Functions were categorised as document-scoped or dataset-scoped functions. Document-scoped functions are operations that modify a document independently of other documents and datasetscoped functions are operations that take into account the whole dataset. Orthogonal to this scope, functions were also separated into cleaning and filtering functions. Cleaning functions aim to remove text considered not part of the main document. Document-scoped cleaning functions can for example target leftover HTML tags. On the other end, dataset-scoped cleaning functions need the whole dataset to calculate a heuristic to determine how to modify each document. For instance, advertisements vary across datasets, making it harder to define a dataset-agnostic classifier for advertisement. Instead, we can index all the lines in a dataset and identify repeated lines on multiple pages as likely advertisements. An example is displayed in Appendix B.2. Filtering functions aim at removing an entire document from the corpus. The reasons for choosing to remove a document completely are diverse: it may be because the document is considered to be of too poor quality, to be too complex to automatically fix or too similar to other examples already present in the corpus. In the latter case, we speak of deduplication. Deduplication of a document is dependent on whether an equivalent document already exists somewhere else in the dataset and is thus necessarily a datasetscope function. The notion of equivalent documents has been explored by Lee et al. (2022). In this case we provide deduplication via metadata (urls, normalised urls) and via text (exact string matching). An exhaustive list of functions is available in B.3.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 173.33050537109375, "coord_origin": "BOTTOMLEFT", "l": 107.24748992919922, "r": 505.6580505371094, "t": 368.76513671875 }, "charspan": [ 0, 1714 ], "page_no": 6 } ], "self_ref": "#/texts/53", "text": "Functions were categorised as document-scoped or dataset-scoped functions. Document-scoped functions are operations that modify a document independently of other documents and datasetscoped functions are operations that take into account the whole dataset. Orthogonal to this scope, functions were also separated into cleaning and filtering functions. Cleaning functions aim to remove text considered not part of the main document. Document-scoped cleaning functions can for example target leftover HTML tags. On the other end, dataset-scoped cleaning functions need the whole dataset to calculate a heuristic to determine how to modify each document. For instance, advertisements vary across datasets, making it harder to define a dataset-agnostic classifier for advertisement. Instead, we can index all the lines in a dataset and identify repeated lines on multiple pages as likely advertisements. An example is displayed in Appendix B.2. Filtering functions aim at removing an entire document from the corpus. The reasons for choosing to remove a document completely are diverse: it may be because the document is considered to be of too poor quality, to be too complex to automatically fix or too similar to other examples already present in the corpus. In the latter case, we speak of deduplication. Deduplication of a document is dependent on whether an equivalent document already exists somewhere else in the dataset and is thus necessarily a datasetscope function. The notion of equivalent documents has been explored by Lee et al. (2022). In this case we provide deduplication via metadata (urls, normalised urls) and via text (exact string matching). An exhaustive list of functions is available in B.3." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As datasets came from heterogeneous sources with different properties, each needs its own set of processing functions to correspond to our definition of natural language documents. In order to support participants in deciding what functions to apply to which, we built and released a streamlit -based visualization tool (figure 2 helps understand the impact of each function, displaying how a document was altered/removed as well as estimated dataset level metrics (quantity of data removed in bytes or samples)). This rapid feedback loop enabled us to update the pipeline consequently in an iterative process to finetune each processing pipelines across datasets and languages with the input of native speakers. A specific example is shared in Appendix B.2. This resulted in 485 non-empty datasets.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 107.1793441772461, "r": 505.6534729003906, "t": 166.79876708984375 }, "charspan": [ 0, 799 ], "page_no": 6 } ], "self_ref": "#/texts/54", "text": "As datasets came from heterogeneous sources with different properties, each needs its own set of processing functions to correspond to our definition of natural language documents. In order to support participants in deciding what functions to apply to which, we built and released a streamlit -based visualization tool (figure 2 helps understand the impact of each function, displaying how a document was altered/removed as well as estimated dataset level metrics (quantity of data removed in bytes or samples)). This rapid feedback loop enabled us to update the pipeline consequently in an iterative process to finetune each processing pipelines across datasets and languages with the input of native speakers. A specific example is shared in Appendix B.2. This resulted in 485 non-empty datasets." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "6", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.71966552734375, "r": 308.5351257324219, "t": 49.4229736328125 }, "charspan": [ 0, 1 ], "page_no": 6 } ], "self_ref": "#/texts/55", "text": "6" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 2: Partial screenshot of the visualization tool. Users can look at how each function in the processing pipeline influenced high-level statistics. Influence on specific samples can be monitored via the same tool, see Appendix B.2", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 602.4551391601562, "coord_origin": "BOTTOMLEFT", "l": 107.43230438232422, "r": 504.0022277832031, "t": 633.9959716796875 }, "charspan": [ 0, 235 ], "page_no": 7 } ], "self_ref": "#/texts/56", "text": "Figure 2: Partial screenshot of the visualization tool. Users can look at how each function in the processing pipeline influenced high-level statistics. Influence on specific samples can be monitored via the same tool, see Appendix B.2" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3 Processing OSCAR", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 567.9241943359375, "coord_origin": "BOTTOMLEFT", "l": 107.35144805908203, "r": 225.0055389404297, "t": 579.6112670898438 }, "charspan": [ 0, 18 ], "page_no": 7 } ], "self_ref": "#/texts/57", "text": "3 Processing OSCAR" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We chose to complement the data obtained at the end of the process described in the previous section with additional Common Crawl-based 8 data motivated by two main reasons. First, given the project's overall goal of providing a trained LLM as a research artifact comparable to previously released ones that have relied extensively on this source, we assessed that not including it would constitute too much of a departure and risk invalidating comparisons. Relatedly, recent work has put a strong emphasis on the quantity of data being a strong factor in a trained model's performance on evaluation tasks (Kaplan et al., 2020; Hoffmann et al., 2022), and we were missing about one third of data in order to optimize our compute budget in this direction. With that in mind, we chose OSCAR version 21.09 (Ortiz Suárez et al., 2020), based on the Common Crawl snapshot of February 2021, to make up the remaining 38% of our final dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 446.4747314453125, "coord_origin": "BOTTOMLEFT", "l": 107.01678466796875, "r": 504.1888427734375, "t": 554.75537109375 }, "charspan": [ 0, 935 ], "page_no": 7 } ], "self_ref": "#/texts/58", "text": "We chose to complement the data obtained at the end of the process described in the previous section with additional Common Crawl-based 8 data motivated by two main reasons. First, given the project's overall goal of providing a trained LLM as a research artifact comparable to previously released ones that have relied extensively on this source, we assessed that not including it would constitute too much of a departure and risk invalidating comparisons. Relatedly, recent work has put a strong emphasis on the quantity of data being a strong factor in a trained model's performance on evaluation tasks (Kaplan et al., 2020; Hoffmann et al., 2022), and we were missing about one third of data in order to optimize our compute budget in this direction. With that in mind, we chose OSCAR version 21.09 (Ortiz Suárez et al., 2020), based on the Common Crawl snapshot of February 2021, to make up the remaining 38% of our final dataset." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "However, crawled data suffers from several known issues. First, we wanted to only select documents written by humans for humans, and exclude machine-generated content e.g. search engine optimization (SEO). Crawled content also over-represents pornographic text across languages (Kreutzer et al., 2022), especially in the form of spam ads. Finally, it contains personal information that may constitute a privacy risk. The present section outlines our approach to mitigating those issues.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 386.21728515625, "coord_origin": "BOTTOMLEFT", "l": 107.12876892089844, "r": 505.65032958984375, "t": 440.12603759765625 }, "charspan": [ 0, 486 ], "page_no": 7 } ], "self_ref": "#/texts/59", "text": "However, crawled data suffers from several known issues. First, we wanted to only select documents written by humans for humans, and exclude machine-generated content e.g. search engine optimization (SEO). Crawled content also over-represents pornographic text across languages (Kreutzer et al., 2022), especially in the form of spam ads. Finally, it contains personal information that may constitute a privacy risk. The present section outlines our approach to mitigating those issues." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.1 Data cleaning and filtering", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 361.1265869140625, "coord_origin": "BOTTOMLEFT", "l": 107.3606948852539, "r": 243.70672607421875, "t": 370.8645935058594 }, "charspan": [ 0, 31 ], "page_no": 7 } ], "self_ref": "#/texts/60", "text": "3.1 Data cleaning and filtering" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Our first approach to addressing the above consists in defining quality indicators for web content. These can then be used to filter out specific pages by defining cutoff thresholds. Extensive descriptions for reproduction are available in appendix C. We filtered out documents with:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 318.9454040527344, "coord_origin": "BOTTOMLEFT", "l": 106.86669158935547, "r": 505.7451171875, "t": 350.8650817871094 }, "charspan": [ 0, 283 ], "page_no": 7 } ], "self_ref": "#/texts/61", "text": "Our first approach to addressing the above consists in defining quality indicators for web content. These can then be used to filter out specific pages by defining cutoff thresholds. Extensive descriptions for reproduction are available in appendix C. We filtered out documents with:" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Too high character repetition or word repetition as a measure of repetitive content.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 298.0774841308594, "coord_origin": "BOTTOMLEFT", "l": 134.97023010253906, "r": 483.9085998535156, "t": 308.407958984375 }, "charspan": [ 0, 86 ], "page_no": 7 } ], "self_ref": "#/texts/62", "text": "· Too high character repetition or word repetition as a measure of repetitive content." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Too high ratios of special characters to remove page code or crawling artifacts.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 282.85308837890625, "coord_origin": "BOTTOMLEFT", "l": 134.86936950683594, "r": 463.3033752441406, "t": 292.74261474609375 }, "charspan": [ 0, 82 ], "page_no": 7 } ], "self_ref": "#/texts/63", "text": "· Too high ratios of special characters to remove page code or crawling artifacts." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Insufficient ratios of closed class words to filter out SEO pages.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 266.70977783203125, "coord_origin": "BOTTOMLEFT", "l": 134.81983947753906, "r": 399.5976867675781, "t": 276.81982421875 }, "charspan": [ 0, 68 ], "page_no": 7 } ], "self_ref": "#/texts/64", "text": "· Insufficient ratios of closed class words to filter out SEO pages." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Too high ratios of flagged words to filter out pornographic spam. We asked contributors to tailor the word list in their language to this criterion (as opposed to generic terms related to sexuality) and to err on the side of high precision.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 229.47686767578125, "coord_origin": "BOTTOMLEFT", "l": 134.82565307617188, "r": 504.00079345703125, "t": 261.60748291015625 }, "charspan": [ 0, 242 ], "page_no": 7 } ], "self_ref": "#/texts/65", "text": "· Too high ratios of flagged words to filter out pornographic spam. We asked contributors to tailor the word list in their language to this criterion (as opposed to generic terms related to sexuality) and to err on the side of high precision." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Too high perplexity values to filter out non-natural language.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 213.66851806640625, "coord_origin": "BOTTOMLEFT", "l": 134.71505737304688, "r": 388.71636962890625, "t": 223.55377197265625 }, "charspan": [ 0, 64 ], "page_no": 7 } ], "self_ref": "#/texts/66", "text": "· Too high perplexity values to filter out non-natural language." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Insufficient number of words , as LLM training requires extensive context sizes.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 198.3580780029297, "coord_origin": "BOTTOMLEFT", "l": 135.04135131835938, "r": 464.89959716796875, "t": 208.28265380859375 }, "charspan": [ 0, 82 ], "page_no": 7 } ], "self_ref": "#/texts/67", "text": "· Insufficient number of words , as LLM training requires extensive context sizes." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The languages that we eventually considered in OSCAR were the languages for which we were able to obtain hyperparameters and the cutoff values for each of these indicators by native speakers. Specifically, we considered Arabic, Basque, Bengali, Catalan, Chinese, English, French, Hindi, Indonesian, Portuguese, Spanish, Urdu, and Vietnamese. The code used for filtering OSCAR, along with the language-specific parameters and cutoff values, are publicly available. We then asked native speakers of each language to use our visualization tool 9 to establish the thresholds for each filter. The percentage of documents removed after applying all these filters is given in Table 1, and the percentage of documents discarded by each filter independently is given in 3.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 100.4892578125, "coord_origin": "BOTTOMLEFT", "l": 107.08076477050781, "r": 505.7413635253906, "t": 187.50372314453125 }, "charspan": [ 0, 763 ], "page_no": 7 } ], "self_ref": "#/texts/68", "text": "The languages that we eventually considered in OSCAR were the languages for which we were able to obtain hyperparameters and the cutoff values for each of these indicators by native speakers. Specifically, we considered Arabic, Basque, Bengali, Catalan, Chinese, English, French, Hindi, Indonesian, Portuguese, Spanish, Urdu, and Vietnamese. The code used for filtering OSCAR, along with the language-specific parameters and cutoff values, are publicly available. We then asked native speakers of each language to use our visualization tool 9 to establish the thresholds for each filter. The percentage of documents removed after applying all these filters is given in Table 1, and the percentage of documents discarded by each filter independently is given in 3." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{8}$https://commoncrawl.org/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 80.0758056640625, "coord_origin": "BOTTOMLEFT", "l": 121.48622131347656, "r": 219.01800537109375, "t": 90.5560302734375 }, "charspan": [ 0, 30 ], "page_no": 7 } ], "self_ref": "#/texts/69", "text": "$^{8}$https://commoncrawl.org/" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{9}$https://hf.co/spaces/huggingface/text-data-filtering", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.969970703125, "coord_origin": "BOTTOMLEFT", "l": 121.45742797851562, "r": 306.87579345703125, "t": 79.2933349609375 }, "charspan": [ 0, 58 ], "page_no": 7 } ], "self_ref": "#/texts/70", "text": "$^{9}$https://hf.co/spaces/huggingface/text-data-filtering" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "7", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960086822509766, "coord_origin": "BOTTOMLEFT", "l": 302.8539123535156, "r": 308.49029541015625, "t": 49.57305908203125 }, "charspan": [ 0, 1 ], "page_no": 7 } ], "self_ref": "#/texts/71", "text": "7" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Table 1: Percentage of documents removed by the filtering per language (ISO 639-1 code).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 675.6669921875, "coord_origin": "BOTTOMLEFT", "l": 124.41507720947266, "r": 486.8222351074219, "t": 685.58251953125 }, "charspan": [ 0, 88 ], "page_no": 8 } ], "self_ref": "#/texts/72", "text": "Table 1: Percentage of documents removed by the filtering per language (ISO 639-1 code)." }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 3: Percentage of documents discarded by each filter independently for 5 languages", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 566.078857421875, "coord_origin": "BOTTOMLEFT", "l": 127.01585388183594, "r": 484.82879638671875, "t": 575.918212890625 }, "charspan": [ 0, 88 ], "page_no": 8 } ], "self_ref": "#/texts/73", "text": "Figure 3: Percentage of documents discarded by each filter independently for 5 languages" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.2 Deduplication", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 531.5645751953125, "coord_origin": "BOTTOMLEFT", "l": 107.01102447509766, "r": 190.201416015625, "t": 541.7603759765625 }, "charspan": [ 0, 17 ], "page_no": 8 } ], "self_ref": "#/texts/74", "text": "3.2 Deduplication" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Data deduplication has become a key tool for language model projects following research showing that it both improves performance on downstream tasks (Lee et al., 2022; Zhang et al., 2021) and decreases memorization of training data (Kandpal et al., 2022). To remove near duplicate documents in OSCAR (which is already exact-deduplicated) we initially used SimHash (Charikar, 2002; Manku et al., 2007), a hashing function that associates to two similar texts hashes with a low Hamming distance, with 6-grams and a Hamming distance threshold of 4. About 0.7% of the documents on average (0.07% ∼ 2.7%) were identified as near duplicates. However, because SimHash is essentially a bag-of-words algorithm, long documents are more likely to end up being similar to each other. In practice, we found false positives among long documents and decided not to discard documents in a same cluster of near-duplicates when they were longer than 6000 characters. Instead, we applied substring deduplication (Lee et al., 2022) based on Suffix Array (Manber and Myers, 1993) as a complementary method that clusters documents sharing a long substring, for documents with more than 6000 characters. We found on average 21.67% (10.61% ∼ 32.30%) of the data (in bytes) being duplicated.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 369.56109619140625, "coord_origin": "BOTTOMLEFT", "l": 107.16806030273438, "r": 504.3442077636719, "t": 520.8411254882812 }, "charspan": [ 0, 1267 ], "page_no": 8 } ], "self_ref": "#/texts/75", "text": "Data deduplication has become a key tool for language model projects following research showing that it both improves performance on downstream tasks (Lee et al., 2022; Zhang et al., 2021) and decreases memorization of training data (Kandpal et al., 2022). To remove near duplicate documents in OSCAR (which is already exact-deduplicated) we initially used SimHash (Charikar, 2002; Manku et al., 2007), a hashing function that associates to two similar texts hashes with a low Hamming distance, with 6-grams and a Hamming distance threshold of 4. About 0.7% of the documents on average (0.07% ∼ 2.7%) were identified as near duplicates. However, because SimHash is essentially a bag-of-words algorithm, long documents are more likely to end up being similar to each other. In practice, we found false positives among long documents and decided not to discard documents in a same cluster of near-duplicates when they were longer than 6000 characters. Instead, we applied substring deduplication (Lee et al., 2022) based on Suffix Array (Manber and Myers, 1993) as a complementary method that clusters documents sharing a long substring, for documents with more than 6000 characters. We found on average 21.67% (10.61% ∼ 32.30%) of the data (in bytes) being duplicated." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.3 Personally identifiable information", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 343.6438293457031, "coord_origin": "BOTTOMLEFT", "l": 107.14714813232422, "r": 278.3106384277344, "t": 353.6887512207031 }, "charspan": [ 0, 39 ], "page_no": 8 } ], "self_ref": "#/texts/76", "text": "3.3 Personally identifiable information" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We used a rule-based approach leveraging regular expressions (Appendix C). The elements redacted were instances of KEY (numeric & alphanumeric identifiers such as phone numbers, credit card numbers, hexadecimal hashes and the like, while skipping instances of years and simple numbers), EMAIL (email addresses), USER (a social media handle) and IP_ADDRESS (an IPv4 or IPv6 address).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 290.26007080078125, "coord_origin": "BOTTOMLEFT", "l": 106.87252807617188, "r": 505.7479248046875, "t": 332.8736267089844 }, "charspan": [ 0, 382 ], "page_no": 8 } ], "self_ref": "#/texts/77", "text": "We used a rule-based approach leveraging regular expressions (Appendix C). The elements redacted were instances of KEY (numeric & alphanumeric identifiers such as phone numbers, credit card numbers, hexadecimal hashes and the like, while skipping instances of years and simple numbers), EMAIL (email addresses), USER (a social media handle) and IP_ADDRESS (an IPv4 or IPv6 address)." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "4 A First look at ROOTS", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 259.9903869628906, "coord_origin": "BOTTOMLEFT", "l": 107.15545654296875, "r": 243.9904022216797, "t": 271.7425537109375 }, "charspan": [ 0, 23 ], "page_no": 8 } ], "self_ref": "#/texts/78", "text": "4 A First look at ROOTS" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The efforts described in the previous sections come together in an assemblage of 1.6 Terabytes of multilingual text. Figure 4 puts that number into context by comparing the sizes of corpora typically used to train large language models. Documentation of the individual components of the corpus can be found in an interactive dataset card deck. In this section, we take initial steps towards further understanding of the corpus through statistical analyses of the aggregated data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 192.2403564453125, "coord_origin": "BOTTOMLEFT", "l": 107.03732299804688, "r": 504.39654541015625, "t": 246.23016357421875 }, "charspan": [ 0, 479 ], "page_no": 8 } ], "self_ref": "#/texts/79", "text": "The efforts described in the previous sections come together in an assemblage of 1.6 Terabytes of multilingual text. Figure 4 puts that number into context by comparing the sizes of corpora typically used to train large language models. Documentation of the individual components of the corpus can be found in an interactive dataset card deck. In this section, we take initial steps towards further understanding of the corpus through statistical analyses of the aggregated data." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "4.1 Natural Languages", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 166.503173828125, "coord_origin": "BOTTOMLEFT", "l": 107.17958068847656, "r": 212.33432006835938, "t": 176.51702880859375 }, "charspan": [ 0, 21 ], "page_no": 8 } ], "self_ref": "#/texts/80", "text": "4.1 Natural Languages" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The constitution of the corpus reflects the crowdsourcing efforts that enabled its creation. It comprises of 46 natural languages spanning 3 macroareas and 9 language families: Afro-Asiatic, AustroAsiatic, Austronesian, Basque, Dravidian, Indo-European, Mande, Niger-Congo, Sino-Tibetan. At 30.03%, English constitutes the largest part of the corpus, followed by Simplified Chinese (16.16%), French (12.9%), Spanish (10.85%), Portuguese (4.91%) and Arabic (4.6%). A more detailed breakdown of the corpus can be found in the appendix and in an online interactive exploration tool$^{10}$,", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 91.41912841796875, "coord_origin": "BOTTOMLEFT", "l": 107.0058364868164, "r": 505.653564453125, "t": 155.61566162109375 }, "charspan": [ 0, 586 ], "page_no": 8 } ], "self_ref": "#/texts/81", "text": "The constitution of the corpus reflects the crowdsourcing efforts that enabled its creation. It comprises of 46 natural languages spanning 3 macroareas and 9 language families: Afro-Asiatic, AustroAsiatic, Austronesian, Basque, Dravidian, Indo-European, Mande, Niger-Congo, Sino-Tibetan. At 30.03%, English constitutes the largest part of the corpus, followed by Simplified Chinese (16.16%), French (12.9%), Spanish (10.85%), Portuguese (4.91%) and Arabic (4.6%). A more detailed breakdown of the corpus can be found in the appendix and in an online interactive exploration tool$^{10}$," }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{10}$https://hf.co/spaces/bigscience-data/corpus-map", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 70.0167236328125, "coord_origin": "BOTTOMLEFT", "l": 119.34101867675781, "r": 298.1865234375, "t": 79.9168701171875 }, "charspan": [ 0, 54 ], "page_no": 8 } ], "self_ref": "#/texts/82", "text": "$^{10}$https://hf.co/spaces/bigscience-data/corpus-map" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "8", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.8652648925781, "r": 308.49029541015625, "t": 49.33624267578125 }, "charspan": [ 0, 1 ], "page_no": 8 } ], "self_ref": "#/texts/83", "text": "8" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 4: A raw size comparison to other corpora used to train large language models. The asterisk next to GPT-3 indicates the fact that the value in question is an estimate computed using the reported number of tokens and the average number of tokens per byte of text that the GPT-2 tokenizer produces on the Pile-CC , Books3 , OWT2 , and Wiki-en subsets of the Pile (Gao et al., 2020)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 492.2463073730469, "coord_origin": "BOTTOMLEFT", "l": 107.27008819580078, "r": 504.2513732910156, "t": 534.5850219726562 }, "charspan": [ 0, 386 ], "page_no": 9 } ], "self_ref": "#/texts/84", "text": "Figure 4: A raw size comparison to other corpora used to train large language models. The asterisk next to GPT-3 indicates the fact that the value in question is an estimate computed using the reported number of tokens and the average number of tokens per byte of text that the GPT-2 tokenizer produces on the Pile-CC , Books3 , OWT2 , and Wiki-en subsets of the Pile (Gao et al., 2020)" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 5: Size in bytes of every document in the corpus per language. The y-axis is in logarithmic scale. Box-and-whisker diagrams illustrate median, the first and third quartiles, whiskers drawn within the 1.5 IQR value and outliers", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 325.1160888671875, "coord_origin": "BOTTOMLEFT", "l": 107.03349304199219, "r": 504.02850341796875, "t": 356.7594299316406 }, "charspan": [ 0, 233 ], "page_no": 9 } ], "self_ref": "#/texts/85", "text": "Figure 5: Size in bytes of every document in the corpus per language. The y-axis is in logarithmic scale. Box-and-whisker diagrams illustrate median, the first and third quartiles, whiskers drawn within the 1.5 IQR value and outliers" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "a screenshot of which is included in figure 1 to depict the byte-distribution of linguistic genera of the Eurasian macroarea subset of the corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 279.4637451171875, "coord_origin": "BOTTOMLEFT", "l": 107.44087982177734, "r": 504.0284118652344, "t": 300.1097412109375 }, "charspan": [ 0, 146 ], "page_no": 9 } ], "self_ref": "#/texts/86", "text": "a screenshot of which is included in figure 1 to depict the byte-distribution of linguistic genera of the Eurasian macroarea subset of the corpus." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In order for the trained model to have an opportunity to learn long dependencies, the training corpus needs to contain long sequences of coherent text. At the same time, the previous post-processing steps only reduced the size of the documents. The median size of a document in our corpus is 1,129 bytes. Figure 5 shows the distribution of document sizes by language. A more detailed breakdown of the size of corpus on an online interactive tool.$^{11}$.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 219.7138671875, "coord_origin": "BOTTOMLEFT", "l": 107.17425537109375, "r": 504.2558898925781, "t": 272.7447509765625 }, "charspan": [ 0, 454 ], "page_no": 9 } ], "self_ref": "#/texts/87", "text": "In order for the trained model to have an opportunity to learn long dependencies, the training corpus needs to contain long sequences of coherent text. At the same time, the previous post-processing steps only reduced the size of the documents. The median size of a document in our corpus is 1,129 bytes. Figure 5 shows the distribution of document sizes by language. A more detailed breakdown of the size of corpus on an online interactive tool.$^{11}$." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The distributions of the filter values for the different filters introduced in Section 3.1 and languages, for the Catalogue, Pseudo-Crawl and OSCAR (filtered) data are available in an online demo$^{12}$. Examples for English are shown in figure 6. The different distributions reflect the diversity of sourcing and filtering of our main components. A notable example is the flagged word filter, for which the distribution for OSCAR is skewed right compared to the catalogue even after filtering.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 159.55859375, "coord_origin": "BOTTOMLEFT", "l": 106.93722534179688, "r": 505.74346923828125, "t": 212.89764404296875 }, "charspan": [ 0, 494 ], "page_no": 9 } ], "self_ref": "#/texts/88", "text": "The distributions of the filter values for the different filters introduced in Section 3.1 and languages, for the Catalogue, Pseudo-Crawl and OSCAR (filtered) data are available in an online demo$^{12}$. Examples for English are shown in figure 6. The different distributions reflect the diversity of sourcing and filtering of our main components. A notable example is the flagged word filter, for which the distribution for OSCAR is skewed right compared to the catalogue even after filtering." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "4.2 Programming Languages", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 133.83482360839844, "coord_origin": "BOTTOMLEFT", "l": 107.1933364868164, "r": 238.58804321289062, "t": 143.71759033203125 }, "charspan": [ 0, 25 ], "page_no": 9 } ], "self_ref": "#/texts/89", "text": "4.2 Programming Languages" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As depicted in the waffle plot in figure 1, the code subset of the corpus spans 13 programming languages, with Java, PHP, and C++ accounting for more than half of all documents.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 102.29308319091797, "coord_origin": "BOTTOMLEFT", "l": 107.02070617675781, "r": 504.1156921386719, "t": 123.03460693359375 }, "charspan": [ 0, 177 ], "page_no": 9 } ], "self_ref": "#/texts/90", "text": "As depicted in the waffle plot in figure 1, the code subset of the corpus spans 13 programming languages, with Java, PHP, and C++ accounting for more than half of all documents." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{11}$https://hf.co/spaces/bigscience-data/document-sizes", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 80.33880615234375, "coord_origin": "BOTTOMLEFT", "l": 119.01773834228516, "r": 311.456787109375, "t": 91.061767578125 }, "charspan": [ 0, 58 ], "page_no": 9 } ], "self_ref": "#/texts/91", "text": "$^{11}$https://hf.co/spaces/bigscience-data/document-sizes" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{12}$https://hf.co/spaces/bigscience-catalogue-lm-data/filter_values_distributions", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 70.06326293945312, "coord_origin": "BOTTOMLEFT", "l": 119.15380859375, "r": 398.82366943359375, "t": 79.9996337890625 }, "charspan": [ 0, 84 ], "page_no": 9 } ], "self_ref": "#/texts/92", "text": "$^{12}$https://hf.co/spaces/bigscience-catalogue-lm-data/filter_values_distributions" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "9", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960086822509766, "coord_origin": "BOTTOMLEFT", "l": 302.5067443847656, "r": 308.49029541015625, "t": 49.33367919921875 }, "charspan": [ 0, 1 ], "page_no": 9 } ], "self_ref": "#/texts/93", "text": "9" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 6: Some distributions of filter values for English. A filter value is the value that the filter gives to a document. These values are generally used to filter out documents that are too low or too high rated and also inform about the composition of the datasets.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 600.04150390625, "coord_origin": "BOTTOMLEFT", "l": 107.46224975585938, "r": 504.0045471191406, "t": 631.2618408203125 }, "charspan": [ 0, 269 ], "page_no": 10 } ], "self_ref": "#/texts/94", "text": "Figure 6: Some distributions of filter values for English. A filter value is the value that the filter gives to a document. These values are generally used to filter out documents that are too low or too high rated and also inform about the composition of the datasets." }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 7: Tokens per byte for each English-language component for tokenizers trained on this corpus (BLOOM), the Pile (GPT-NeoX 20B) and C4 (T5). Lower values mean the component (X axis) is more similar in aggregate to the compared training corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 407.1468200683594, "coord_origin": "BOTTOMLEFT", "l": 107.3069839477539, "r": 503.998046875, "t": 438.5857849121094 }, "charspan": [ 0, 248 ], "page_no": 10 } ], "self_ref": "#/texts/95", "text": "Figure 7: Tokens per byte for each English-language component for tokenizers trained on this corpus (BLOOM), the Pile (GPT-NeoX 20B) and C4 (T5). Lower values mean the component (X axis) is more similar in aggregate to the compared training corpus." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Configuration and test files are abundant in most GitHub repositories but not as interesting for code modeling. To that end, we use a heuristic whose first step examines the first 5 lines of a file for the presence of keywords such as \"configuration file\" or \"test file\". Failing that, the second step is to see whether the occurrence of the literals config and test in a given file exceeds 5% of the total number of lines of that file. We find that 5.23% of the data consists of configuration files and 7.88% of test files.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 318.236083984375, "coord_origin": "BOTTOMLEFT", "l": 107.19539642333984, "r": 504.82196044921875, "t": 382.2506103515625 }, "charspan": [ 0, 524 ], "page_no": 10 } ], "self_ref": "#/texts/96", "text": "Configuration and test files are abundant in most GitHub repositories but not as interesting for code modeling. To that end, we use a heuristic whose first step examines the first 5 lines of a file for the presence of keywords such as \"configuration file\" or \"test file\". Failing that, the second step is to see whether the occurrence of the literals config and test in a given file exceeds 5% of the total number of lines of that file. We find that 5.23% of the data consists of configuration files and 7.88% of test files." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Allamanis (2019) and Lopes et al. (2017) highlight the large fraction of near-duplicates present in code datasets and how they can inflate performance metrics. Exact match deduplication alone can miss a fair amount of near-duplicates. To detect them, we first compute the MinHash of all documents, then create a Locality Sensitive Hashing (LSH) index between files to find the duplicate clusters in linear time. We additionally evaluate the Jaccard similarities within duplicate clusters to remove some false positives. We find 10.9M duplicate files in the clusters and 4.1M unique files: almost 32% of the data consists of near-duplicates. Syntax checkers 13 are used to validate 500K samples of Python and PHP code. We find that only 1% of the Python data and 2% of the PHP files do not pass the syntax check.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 214.1485595703125, "coord_origin": "BOTTOMLEFT", "l": 107.21968841552734, "r": 505.24127197265625, "t": 311.57904052734375 }, "charspan": [ 0, 811 ], "page_no": 10 } ], "self_ref": "#/texts/97", "text": "Allamanis (2019) and Lopes et al. (2017) highlight the large fraction of near-duplicates present in code datasets and how they can inflate performance metrics. Exact match deduplication alone can miss a fair amount of near-duplicates. To detect them, we first compute the MinHash of all documents, then create a Locality Sensitive Hashing (LSH) index between files to find the duplicate clusters in linear time. We additionally evaluate the Jaccard similarities within duplicate clusters to remove some false positives. We find 10.9M duplicate files in the clusters and 4.1M unique files: almost 32% of the data consists of near-duplicates. Syntax checkers 13 are used to validate 500K samples of Python and PHP code. We find that only 1% of the Python data and 2% of the PHP files do not pass the syntax check." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "4.3 Tokenizer analysis of the component datasets", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 188.19244384765625, "coord_origin": "BOTTOMLEFT", "l": 107.28107452392578, "r": 321.77301025390625, "t": 198.179931640625 }, "charspan": [ 0, 48 ], "page_no": 10 } ], "self_ref": "#/texts/98", "text": "4.3 Tokenizer analysis of the component datasets" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "A tokenizer trained on a dataset can be used as a proxy for its content (Gao et al., 2020). The relevant metric is the number of tokens produced for a byte of natural language. The more different the training corpus from the tokenized corpus, the more tokens will be produced as the tokenizer is forced to divide natural text in more numerous, more general, smaller tokens. This property has allowed us to spot errors associated with outlier values, such as incorrectly classified languages, or crawling error. In the following analysis, we use it in two ways: first, we can use tokenizers trained on different corpora to see how ours differs from them; and second, we can use a tokenizer trained on this corpus to assess which components are outliers. We exclude outliers smaller than 5 documents.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 91.59907531738281, "coord_origin": "BOTTOMLEFT", "l": 107.17161560058594, "r": 504.039306640625, "t": 177.33026123046875 }, "charspan": [ 0, 798 ], "page_no": 10 } ], "self_ref": "#/texts/99", "text": "A tokenizer trained on a dataset can be used as a proxy for its content (Gao et al., 2020). The relevant metric is the number of tokens produced for a byte of natural language. The more different the training corpus from the tokenized corpus, the more tokens will be produced as the tokenizer is forced to divide natural text in more numerous, more general, smaller tokens. This property has allowed us to spot errors associated with outlier values, such as incorrectly classified languages, or crawling error. In the following analysis, we use it in two ways: first, we can use tokenizers trained on different corpora to see how ours differs from them; and second, we can use a tokenizer trained on this corpus to assess which components are outliers. We exclude outliers smaller than 5 documents." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{13}$py_compile for Python and the -l flag for PHP", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.5452880859375, "coord_origin": "BOTTOMLEFT", "l": 119.4175033569336, "r": 300.5633239746094, "t": 80.13006591796875 }, "charspan": [ 0, 52 ], "page_no": 10 } ], "self_ref": "#/texts/100", "text": "$^{13}$py_compile for Python and the -l flag for PHP" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "10", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 301.01898193359375, "r": 311.2897033691406, "t": 49.3214111328125 }, "charspan": [ 0, 2 ], "page_no": 10 } ], "self_ref": "#/texts/101", "text": "10" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 8: Tokens per byte for each French, Simplified Chinese, and Arabic component for tokenizers trained on this corpus. Lower values mean the component (X axis) is more similar in aggregate to the rest of the corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 596.943115234375, "coord_origin": "BOTTOMLEFT", "l": 107.4345474243164, "r": 504.0013122558594, "t": 628.757080078125 }, "charspan": [ 0, 219 ], "page_no": 11 } ], "self_ref": "#/texts/102", "text": "Figure 8: Tokens per byte for each French, Simplified Chinese, and Arabic component for tokenizers trained on this corpus. Lower values mean the component (X axis) is more similar in aggregate to the rest of the corpus." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Figure 7 shows the tokens-per-byte measurement on English component datasets for the BLOOM tokenizer, trained on this corpus, the GPT-NeoX 20B tokenizer (Black et al., 2022), trained on the Pile, and the T5 tokenizer (Raffel et al., 2020), trained on C4. Those tokenizers may differ in algorithms and/or vocabulary size, but we won't be directly comparing them to each other.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 532.2470703125, "coord_origin": "BOTTOMLEFT", "l": 107.265869140625, "r": 505.2414245605469, "t": 574.7288208007812 }, "charspan": [ 0, 375 ], "page_no": 11 } ], "self_ref": "#/texts/103", "text": "Figure 7 shows the tokens-per-byte measurement on English component datasets for the BLOOM tokenizer, trained on this corpus, the GPT-NeoX 20B tokenizer (Black et al., 2022), trained on the Pile, and the T5 tokenizer (Raffel et al., 2020), trained on C4. Those tokenizers may differ in algorithms and/or vocabulary size, but we won't be directly comparing them to each other." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The figure is ordered by BLOOM tokenizer token-per-byte values, which shows that the ordering is very similar for BLOOM and GPT-NeoX. However, it shows several bumps for T5: component datasets that are out of domain in C4 but not our corpus, for example technical and academic datasets such as s2orc or royal_society_corpus , domains absent from C4's Common Crawl-sourced data. Other such datasets include global_voices , which contains news about non-English-speaking regions including quotes in the original languages and no_code_stackexchange , which contains forums which, although in English, may be dedicated to technical matters, foreign languages, or very specific domains. Both are similar to our corpus but not to the Pile or C4.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 439.05902099609375, "coord_origin": "BOTTOMLEFT", "l": 107.1494140625, "r": 504.35186767578125, "t": 525.46142578125 }, "charspan": [ 0, 739 ], "page_no": 11 } ], "self_ref": "#/texts/104", "text": "The figure is ordered by BLOOM tokenizer token-per-byte values, which shows that the ordering is very similar for BLOOM and GPT-NeoX. However, it shows several bumps for T5: component datasets that are out of domain in C4 but not our corpus, for example technical and academic datasets such as s2orc or royal_society_corpus , domains absent from C4's Common Crawl-sourced data. Other such datasets include global_voices , which contains news about non-English-speaking regions including quotes in the original languages and no_code_stackexchange , which contains forums which, although in English, may be dedicated to technical matters, foreign languages, or very specific domains. Both are similar to our corpus but not to the Pile or C4." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Figure 8 additionally shows BLOOM fertilities for Simplified Chinese, French and Arabic components. Outlier, high-fertility components, e.g. datasets that differ from the rest of our corpus, tend to be the same for all languages. project_gutenberg contains old books with their original formatting (for example, \"***********\" to denote page ends). wiktionary contains definitions of words in foreign languages. wikiversity contains technical terms and L A T E X. wikivoyage contains tables formatted as text. Forums may contain the user and date information of the message, as well as internet slang or emoji. arabench is spoken Arabic, and habibi is classical Arabic with more diacritics than modern. We deem most of those deviations acceptable to represent the diversity of uses of text, which tokenizer analysis is able to surface from the rest of the dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 335.80621337890625, "coord_origin": "BOTTOMLEFT", "l": 107.18116760253906, "r": 505.6535949707031, "t": 432.8155212402344 }, "charspan": [ 0, 863 ], "page_no": 11 } ], "self_ref": "#/texts/105", "text": "Figure 8 additionally shows BLOOM fertilities for Simplified Chinese, French and Arabic components. Outlier, high-fertility components, e.g. datasets that differ from the rest of our corpus, tend to be the same for all languages. project_gutenberg contains old books with their original formatting (for example, \"***********\" to denote page ends). wiktionary contains definitions of words in foreign languages. wikiversity contains technical terms and L A T E X. wikivoyage contains tables formatted as text. Forums may contain the user and date information of the message, as well as internet slang or emoji. arabench is spoken Arabic, and habibi is classical Arabic with more diacritics than modern. We deem most of those deviations acceptable to represent the diversity of uses of text, which tokenizer analysis is able to surface from the rest of the dataset." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5 Conclusion", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 307.2593688964844, "coord_origin": "BOTTOMLEFT", "l": 107.57479858398438, "r": 183.06671142578125, "t": 318.96966552734375 }, "charspan": [ 0, 12 ], "page_no": 11 } ], "self_ref": "#/texts/106", "text": "5 Conclusion" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We have presented ROOTS, a massive multilingual corpus that was the result of an international collaboration between multidisciplinary researchers studying large language models. The efforts to put the corpus together were value-driven and prompted by a data-first approach to training the BLOOM model. We further release the tooling developed throughout the project, and are currently implementing a release strategy that is informed by both the licensing and governance needs of every data source for the corpus itself. We hope this paves the way toward a more reflected use of the data that makes its way into large language models.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 218.97314453125, "coord_origin": "BOTTOMLEFT", "l": 107.10443878173828, "r": 504.349609375, "t": 294.1182556152344 }, "charspan": [ 0, 635 ], "page_no": 11 } ], "self_ref": "#/texts/107", "text": "We have presented ROOTS, a massive multilingual corpus that was the result of an international collaboration between multidisciplinary researchers studying large language models. The efforts to put the corpus together were value-driven and prompted by a data-first approach to training the BLOOM model. We further release the tooling developed throughout the project, and are currently implementing a release strategy that is informed by both the licensing and governance needs of every data source for the corpus itself. We hope this paves the way toward a more reflected use of the data that makes its way into large language models." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Ethical Considerations and Broader Impacts Statement", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 190.63035583496094, "coord_origin": "BOTTOMLEFT", "l": 107.60047149658203, "r": 391.05133056640625, "t": 202.2239990234375 }, "charspan": [ 0, 52 ], "page_no": 11 } ], "self_ref": "#/texts/108", "text": "Ethical Considerations and Broader Impacts Statement" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As discussed in Section 1, the BigScience Research Workshop was conceived as a collaborative and value-driven endeavor from the start. This approach shaped many of the decisions described in this paper, spurring many contextual discussions and consensus-seeking on how to articulate the project's core values, those of the contributors to the data efforts, and considerations of social impact on the people directly and indirectly impacted. Of particular relevance were the data release and governance strategy, the choice to center human selection of data while still using OSCAR web-crawled for a significant section of the corpus, and the tools we developed to manage the risks of the latter (including regarding privacy). Each of these were the occasion of moral exercises and technical contributions that we believe were useful and required, and each will require further research and progress. We provide a more detailed discussion of these aspects of our work in Appendix A.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.33819580078125, "coord_origin": "BOTTOMLEFT", "l": 106.92864990234375, "r": 504.2185363769531, "t": 177.41448974609375 }, "charspan": [ 0, 981 ], "page_no": 11 } ], "self_ref": "#/texts/109", "text": "As discussed in Section 1, the BigScience Research Workshop was conceived as a collaborative and value-driven endeavor from the start. This approach shaped many of the decisions described in this paper, spurring many contextual discussions and consensus-seeking on how to articulate the project's core values, those of the contributors to the data efforts, and considerations of social impact on the people directly and indirectly impacted. Of particular relevance were the data release and governance strategy, the choice to center human selection of data while still using OSCAR web-crawled for a significant section of the corpus, and the tools we developed to manage the risks of the latter (including regarding privacy). Each of these were the occasion of moral exercises and technical contributions that we believe were useful and required, and each will require further research and progress. We provide a more detailed discussion of these aspects of our work in Appendix A." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "11", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.9881286621094, "r": 310.9815673828125, "t": 49.354248046875 }, "charspan": [ 0, 2 ], "page_no": 11 } ], "self_ref": "#/texts/110", "text": "11" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Acknowledgements", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.2732543945312, "coord_origin": "BOTTOMLEFT", "l": 107.2573013305664, "r": 206.83364868164062, "t": 718.769287109375 }, "charspan": [ 0, 16 ], "page_no": 12 } ], "self_ref": "#/texts/111", "text": "Acknowledgements" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "BigScience. This work was pursued as part of the BigScience research workshop, an effort to collaboratively build a very large multilingual neural network language model and a very large multilingual text dataset. This effort gathered 1000+ reasearchers from 60 countries and from more than 250 institutions.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 652.5321044921875, "coord_origin": "BOTTOMLEFT", "l": 107.15486907958984, "r": 504.0410461425781, "t": 694.7559814453125 }, "charspan": [ 0, 308 ], "page_no": 12 } ], "self_ref": "#/texts/112", "text": "BigScience. This work was pursued as part of the BigScience research workshop, an effort to collaboratively build a very large multilingual neural network language model and a very large multilingual text dataset. This effort gathered 1000+ reasearchers from 60 countries and from more than 250 institutions." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Compute. The BigScience Workshop was granted access to the HPC resources of the Institut du développement et des ressources en informatique scientifique (IDRIS) du Centre national de la recherche scientifique (CNRS) under the allocation 2021-A0101012475 made by Grand équipement national de calcul intensif (GENCI). Model training ran on the Jean-Zay cluster of IDRIS, and we thank the IDRIS team for their responsive support throughout the project, in particular Rémi Lacroix.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 592.4578857421875, "coord_origin": "BOTTOMLEFT", "l": 107.22928619384766, "r": 505.6766662597656, "t": 645.5387573242188 }, "charspan": [ 0, 477 ], "page_no": 12 } ], "self_ref": "#/texts/113", "text": "Compute. The BigScience Workshop was granted access to the HPC resources of the Institut du développement et des ressources en informatique scientifique (IDRIS) du Centre national de la recherche scientifique (CNRS) under the allocation 2021-A0101012475 made by Grand équipement national de calcul intensif (GENCI). Model training ran on the Jean-Zay cluster of IDRIS, and we thank the IDRIS team for their responsive support throughout the project, in particular Rémi Lacroix." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "References", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 564.2543334960938, "coord_origin": "BOTTOMLEFT", "l": 107.55571746826172, "r": 163.65428161621094, "t": 575.487060546875 }, "charspan": [ 0, 10 ], "page_no": 12 } ], "self_ref": "#/texts/114", "text": "References" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Abadji, J., P. J. Ortiz Suárez, L. Romary, and B. Sagot (2021, July). Ungoliant: An Optimized Pipeline for the Generation of a Very Large-Scale Multilingual Web Corpus. In CMLC 2021 - 9th Workshop on Challenges in the Management of Large Corpora , Limerick / Virtual, Ireland.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 525.6007080078125, "coord_origin": "BOTTOMLEFT", "l": 107.77008819580078, "r": 504.00189208984375, "t": 556.914306640625 }, "charspan": [ 0, 276 ], "page_no": 12 } ], "self_ref": "#/texts/115", "text": "Abadji, J., P. J. Ortiz Suárez, L. Romary, and B. Sagot (2021, July). Ungoliant: An Optimized Pipeline for the Generation of a Very Large-Scale Multilingual Web Corpus. In CMLC 2021 - 9th Workshop on Challenges in the Management of Large Corpora , Limerick / Virtual, Ireland." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Abdelali, A., F. Guzman, H. Sajjad, and S. Vogel (2014, may). The amara corpus: Building parallel language resources for the educational domain. In N. C. C. Chair), K. Choukri, T. Declerck, H. Loftsson, B. Maegaard, J. Mariani, A. Moreno, J. Odijk, and S. Piperidis (Eds.), Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14) , Reykjavik, Iceland. European Language Resources Association (ELRA).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 464.0589294433594, "coord_origin": "BOTTOMLEFT", "l": 107.4380874633789, "r": 505.51690673828125, "t": 517.4777221679688 }, "charspan": [ 0, 440 ], "page_no": 12 } ], "self_ref": "#/texts/116", "text": "Abdelali, A., F. Guzman, H. Sajjad, and S. Vogel (2014, may). The amara corpus: Building parallel language resources for the educational domain. In N. C. C. Chair), K. Choukri, T. Declerck, H. Loftsson, B. Maegaard, J. Mariani, A. Moreno, J. Odijk, and S. Piperidis (Eds.), Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC'14) , Reykjavik, Iceland. European Language Resources Association (ELRA)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Adelani, D. I., J. Abbott, G. Neubig, D. D'souza, J. Kreutzer, C. Lignos, C. Palen-Michel, H. Buzaaba, S. Rijhwani, S. Ruder, S. Mayhew, I. A. Azime, S. H. Muhammad, C. C. Emezue, J. NakatumbaNabende, P. Ogayo, A. Anuoluwapo, C. Gitau, D. Mbaye, J. Alabi, S. M. Yimam, T. R. Gwadabe, I. Ezeani, R. A. Niyongabo, J. Mukiibi, V. Otiende, I. Orife, D. David, S. Ngom, T. Adewumi, P. Rayson, M. Adeyemi, G. Muriuki, E. Anebi, C. Chukwuneke, N. Odu, E. P. Wairagala, S. Oyerinde, C. Siro, T. S. Bateesa, T. Oloyede, Y. Wambui, V. Akinode, D. Nabagereka, M. Katusiime, A. Awokoya, M. MBOUP, D. Gebreyohannes, H. Tilaye, K. Nwaike, D. Wolde, A. Faye, B. Sibanda, O. Ahia, B. F. P. Dossou, K. Ogueji, T. I. DIOP, A. Diallo, A. Akinfaderin, T. Marengereke, and S. Osei (2021). MasakhaNER: Named entity recognition for African languages. Transactions of the Association for Computational Linguistics 9 , 1116-1131.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 348.2152099609375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.6543884277344, "t": 456.0801086425781 }, "charspan": [ 0, 904 ], "page_no": 12 } ], "self_ref": "#/texts/117", "text": "Adelani, D. I., J. Abbott, G. Neubig, D. D'souza, J. Kreutzer, C. Lignos, C. Palen-Michel, H. Buzaaba, S. Rijhwani, S. Ruder, S. Mayhew, I. A. Azime, S. H. Muhammad, C. C. Emezue, J. NakatumbaNabende, P. Ogayo, A. Anuoluwapo, C. Gitau, D. Mbaye, J. Alabi, S. M. Yimam, T. R. Gwadabe, I. Ezeani, R. A. Niyongabo, J. Mukiibi, V. Otiende, I. Orife, D. David, S. Ngom, T. Adewumi, P. Rayson, M. Adeyemi, G. Muriuki, E. Anebi, C. Chukwuneke, N. Odu, E. P. Wairagala, S. Oyerinde, C. Siro, T. S. Bateesa, T. Oloyede, Y. Wambui, V. Akinode, D. Nabagereka, M. Katusiime, A. Awokoya, M. MBOUP, D. Gebreyohannes, H. Tilaye, K. Nwaike, D. Wolde, A. Faye, B. Sibanda, O. Ahia, B. F. P. Dossou, K. Ogueji, T. I. DIOP, A. Diallo, A. Akinfaderin, T. Marengereke, and S. Osei (2021). MasakhaNER: Named entity recognition for African languages. Transactions of the Association for Computational Linguistics 9 , 1116-1131." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Aghajanyan, A., D. Okhonko, M. Lewis, M. Joshi, H. Xu, G. Ghosh, and L. Zettlemoyer (2022). HTLM: Hyper-text pre-training and prompting of language models. In International Conference on Learning Representations .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 308.1875915527344, "coord_origin": "BOTTOMLEFT", "l": 107.48568725585938, "r": 505.744873046875, "t": 340.38507080078125 }, "charspan": [ 0, 213 ], "page_no": 12 } ], "self_ref": "#/texts/118", "text": "Aghajanyan, A., D. Okhonko, M. Lewis, M. Joshi, H. Xu, G. Ghosh, and L. Zettlemoyer (2022). HTLM: Hyper-text pre-training and prompting of language models. In International Conference on Learning Representations ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Agi'c, Ž. and I. Vuli'c (2019, July). JW300: A wide-coverage parallel corpus for low-resource languages. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics , Florence, Italy, pp. 3204-3210. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 267.55694580078125, "coord_origin": "BOTTOMLEFT", "l": 107.46460723876953, "r": 504.4248962402344, "t": 300.0082702636719 }, "charspan": [ 0, 272 ], "page_no": 12 } ], "self_ref": "#/texts/119", "text": "Agi'c, Ž. and I. Vuli'c (2019, July). JW300: A wide-coverage parallel corpus for low-resource languages. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics , Florence, Italy, pp. 3204-3210. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Allamanis, M. (2019). The adverse effects of code duplication in machine learning models of code. In Proceedings of the 2019 ACM SIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software , pp. 143-153.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 228.3060760498047, "coord_origin": "BOTTOMLEFT", "l": 107.50582885742188, "r": 505.7442626953125, "t": 259.65289306640625 }, "charspan": [ 0, 248 ], "page_no": 12 } ], "self_ref": "#/texts/120", "text": "Allamanis, M. (2019). The adverse effects of code duplication in machine learning models of code. In Proceedings of the 2019 ACM SIGPLAN International Symposium on New Ideas, New Paradigms, and Reflections on Programming and Software , pp. 143-153." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Alrabiah, M., A. Alsalman, and E. Atwell (2013, 01). The design and construction of the 50 million words ksucca king saud university corpus of classical arabic.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 199.60108947753906, "coord_origin": "BOTTOMLEFT", "l": 107.42101287841797, "r": 504.0013732910156, "t": 220.18280029296875 }, "charspan": [ 0, 160 ], "page_no": 12 } ], "self_ref": "#/texts/121", "text": "Alrabiah, M., A. Alsalman, and E. Atwell (2013, 01). The design and construction of the 50 million words ksucca king saud university corpus of classical arabic." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Aly, M. and A. Atiya (2013, August). LABR: A large scale Arabic book reviews dataset. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) , Sofia, Bulgaria, pp. 494-498. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 159.58038330078125, "coord_origin": "BOTTOMLEFT", "l": 107.53243255615234, "r": 504.2693176269531, "t": 191.361083984375 }, "charspan": [ 0, 276 ], "page_no": 12 } ], "self_ref": "#/texts/122", "text": "Aly, M. and A. Atiya (2013, August). LABR: A large scale Arabic book reviews dataset. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers) , Sofia, Bulgaria, pp. 494-498. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Alyafeai, Z., M. Masoud, M. Ghaleb, and M. S. AlShaibani (2021). Masader: Metadata sourcing for arabic text and speech data resources. CoRR abs/2110.06744 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 131.24029541015625, "coord_origin": "BOTTOMLEFT", "l": 107.41636657714844, "r": 504.68707275390625, "t": 152.22198486328125 }, "charspan": [ 0, 156 ], "page_no": 12 } ], "self_ref": "#/texts/123", "text": "Alyafeai, Z., M. Masoud, M. Ghaleb, and M. S. AlShaibani (2021). Masader: Metadata sourcing for arabic text and speech data resources. CoRR abs/2110.06744 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Armengol-Estapé, J., C. P. Carrino, C. Rodriguez-Penagos, O. de Gibert Bonet, C. Armentano-Oller, A. Gonzalez-Agirre, M. Melero, and M. Villegas (2021, August). Are multilingual models the best choice for moderately under-resourced languages? A comprehensive assessment for Catalan. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021 , Online, pp. 4933-4946. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.71197509765625, "coord_origin": "BOTTOMLEFT", "l": 107.2345962524414, "r": 505.74603271484375, "t": 122.9271240234375 }, "charspan": [ 0, 428 ], "page_no": 12 } ], "self_ref": "#/texts/124", "text": "Armengol-Estapé, J., C. P. Carrino, C. Rodriguez-Penagos, O. de Gibert Bonet, C. Armentano-Oller, A. Gonzalez-Agirre, M. Melero, and M. Villegas (2021, August). Are multilingual models the best choice for moderately under-resourced languages? A comprehensive assessment for Catalan. In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021 , Online, pp. 4933-4946. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "12", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.97027587890625, "r": 311.0296630859375, "t": 49.62054443359375 }, "charspan": [ 0, 2 ], "page_no": 12 } ], "self_ref": "#/texts/125", "text": "12" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Artetxe, M., I. Aldabe, R. Agerri, O. Perez-de Viñaspre, and A. Soroa (2022). Does corpus quality really matter for low-resource languages?", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.97607421875, "coord_origin": "BOTTOMLEFT", "l": 107.2368392944336, "r": 504.35101318359375, "t": 717.655517578125 }, "charspan": [ 0, 139 ], "page_no": 13 } ], "self_ref": "#/texts/126", "text": "Artetxe, M., I. Aldabe, R. Agerri, O. Perez-de Viñaspre, and A. Soroa (2022). Does corpus quality really matter for low-resource languages?" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ashari, A. (2018). Indonesian news articles published at 2017.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 678.51953125, "coord_origin": "BOTTOMLEFT", "l": 107.2130126953125, "r": 357.8719787597656, "t": 688.1885986328125 }, "charspan": [ 0, 62 ], "page_no": 13 } ], "self_ref": "#/texts/127", "text": "Ashari, A. (2018). Indonesian news articles published at 2017." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Bandy, J. and N. Vincent (2021). Addressing\" documentation debt\" in machine learning research: A retrospective datasheet for bookcorpus. arXiv preprint arXiv:2105.05241 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 649.5961303710938, "coord_origin": "BOTTOMLEFT", "l": 107.81269836425781, "r": 504.5561218261719, "t": 670.2788696289062 }, "charspan": [ 0, 170 ], "page_no": 13 } ], "self_ref": "#/texts/128", "text": "Bandy, J. and N. Vincent (2021). Addressing\" documentation debt\" in machine learning research: A retrospective datasheet for bookcorpus. arXiv preprint arXiv:2105.05241 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Belinkov, Y., A. Magidow, A. Barrón-Cedeño, A. Shmidman, and M. Romanov (2019). Studying the history of the arabic language: language technology and a large-scale historical corpus. Language Resources and Evaluation 53 (4), 771-805.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 609.6190795898438, "coord_origin": "BOTTOMLEFT", "l": 107.88778686523438, "r": 504.41455078125, "t": 641.0530395507812 }, "charspan": [ 0, 232 ], "page_no": 13 } ], "self_ref": "#/texts/129", "text": "Belinkov, Y., A. Magidow, A. Barrón-Cedeño, A. Shmidman, and M. Romanov (2019). Studying the history of the arabic language: language technology and a large-scale historical corpus. Language Resources and Evaluation 53 (4), 771-805." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Bender, E. M., T. Gebru, A. McMillan-Major, and S. Shmitchell (2021). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency , FAccT '21, New York, NY, USA, pp. 610-623. Association for Computing Machinery.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 557.8916015625, "coord_origin": "BOTTOMLEFT", "l": 107.81028747558594, "r": 505.73876953125, "t": 600.8572998046875 }, "charspan": [ 0, 309 ], "page_no": 13 } ], "self_ref": "#/texts/130", "text": "Bender, E. M., T. Gebru, A. McMillan-Major, and S. Shmitchell (2021). On the dangers of stochastic parrots: Can language models be too big? In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency , FAccT '21, New York, NY, USA, pp. 610-623. Association for Computing Machinery." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Biderman, S., K. Bicheno, and L. Gao (2022). Datasheet for the pile. arXiv preprint arXiv:2201.07311 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 529.5631103515625, "coord_origin": "BOTTOMLEFT", "l": 107.61126708984375, "r": 504.7915344238281, "t": 550.22314453125 }, "charspan": [ 0, 102 ], "page_no": 13 } ], "self_ref": "#/texts/131", "text": "Biderman, S., K. Bicheno, and L. Gao (2022). Datasheet for the pile. arXiv preprint arXiv:2201.07311 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "BigScience Workshop (2022). Bloom (revision 4ab0472).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 510.9230651855469, "coord_origin": "BOTTOMLEFT", "l": 107.43785858154297, "r": 339.1024169921875, "t": 520.9296875 }, "charspan": [ 0, 53 ], "page_no": 13 } ], "self_ref": "#/texts/132", "text": "BigScience Workshop (2022). Bloom (revision 4ab0472)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Bird, S., E. Klein, and E. Loper (2009). Natural Language Processing with Python: Analyzing Text with the Natural Language Toolkit . Beijing: O'Reilly.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 482.0586242675781, "coord_origin": "BOTTOMLEFT", "l": 107.8008041381836, "r": 504.54852294921875, "t": 502.8025207519531 }, "charspan": [ 0, 151 ], "page_no": 13 } ], "self_ref": "#/texts/133", "text": "Bird, S., E. Klein, and E. Loper (2009). Natural Language Processing with Python: Analyzing Text with the Natural Language Toolkit . Beijing: O'Reilly." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Birhane, A., P. Kalluri, D. Card, W. Agnew, R. Dotan, and M. Bao (2021). The values encoded in machine learning research. ArXiv abs/2106.15590 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 452.9429626464844, "coord_origin": "BOTTOMLEFT", "l": 107.70643615722656, "r": 504.0847473144531, "t": 473.77532958984375 }, "charspan": [ 0, 144 ], "page_no": 13 } ], "self_ref": "#/texts/134", "text": "Birhane, A., P. Kalluri, D. Card, W. Agnew, R. Dotan, and M. Bao (2021). The values encoded in machine learning research. ArXiv abs/2106.15590 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Black, S., S. Biderman, E. Hallahan, Q. Anthony, L. Gao, L. Golding, H. He, C. Leahy, K. McDonell, J. Phang, et al. (2022). Gpt-neox-20b: An open-source autoregressive language model. In Proceedings of BigScience Episode \\ # 5-Workshop on Challenges & Perspectives in Creating Large Language Models , pp. 95-136.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 401.41119384765625, "coord_origin": "BOTTOMLEFT", "l": 107.47306823730469, "r": 505.2412414550781, "t": 445.0248107910156 }, "charspan": [ 0, 312 ], "page_no": 13 } ], "self_ref": "#/texts/135", "text": "Black, S., S. Biderman, E. Hallahan, Q. Anthony, L. Gao, L. Golding, H. He, C. Leahy, K. McDonell, J. Phang, et al. (2022). Gpt-neox-20b: An open-source autoregressive language model. In Proceedings of BigScience Episode \\ # 5-Workshop on Challenges & Perspectives in Creating Large Language Models , pp. 95-136." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Bondi, E., L. Xu, D. Acosta-Navas, and J. A. Killian (2021). Envisioning communities: A participatory approach towards AI for social good. In M. Fourcade, B. Kuipers, S. Lazar, and D. K. Mulligan (Eds.), AIES '21: AAAI/ACM Conference on AI, Ethics, and Society, Virtual Event, USA, May 19-21, 2021 , pp. 425-436. ACM.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 350.496337890625, "coord_origin": "BOTTOMLEFT", "l": 107.65503692626953, "r": 505.7463073730469, "t": 393.6221008300781 }, "charspan": [ 0, 317 ], "page_no": 13 } ], "self_ref": "#/texts/136", "text": "Bondi, E., L. Xu, D. Acosta-Navas, and J. A. Killian (2021). Envisioning communities: A participatory approach towards AI for social good. In M. Fourcade, B. Kuipers, S. Lazar, and D. K. Mulligan (Eds.), AIES '21: AAAI/ACM Conference on AI, Ethics, and Society, Virtual Event, USA, May 19-21, 2021 , pp. 425-436. ACM." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Brown, T. B., B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert-Voss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Amodei (2020). Language models are few-shot learners. In H. Larochelle, M. Ranzato, R. Hadsell, M. Balcan, and H. Lin (Eds.), Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual .", "parent": { "$ref": "#/groups/1" }, "prov": [ { "bbox": { "b": 267.5760803222656, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.24908447265625, "t": 342.67010498046875 }, "charspan": [ 0, 643 ], "page_no": 13 } ], "self_ref": "#/texts/137", "text": "Brown, T. B., B. Mann, N. Ryder, M. Subbiah, J. Kaplan, P. Dhariwal, A. Neelakantan, P. Shyam, G. Sastry, A. Askell, S. Agarwal, A. Herbert-Voss, G. Krueger, T. Henighan, R. Child, A. Ramesh, D. M. Ziegler, J. Wu, C. Winter, C. Hesse, M. Chen, E. Sigler, M. Litwin, S. Gray, B. Chess, J. Clark, C. Berner, S. McCandlish, A. Radford, I. Sutskever, and D. Amodei (2020). Language models are few-shot learners. In H. Larochelle, M. Ranzato, R. Hadsell, M. Balcan, and H. Lin (Eds.), Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Budiono, H. Riza, and C. Hakim (2009, August). Resource report: Building parallel text corpora for multi-domain translation system. In Proceedings of the 7th Workshop on Asian Language Resources (ALR7) , Suntec, Singapore, pp. 92-95. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 227.5013427734375, "coord_origin": "BOTTOMLEFT", "l": 107.59933471679688, "r": 504.0403747558594, "t": 258.9801025390625 }, "charspan": [ 0, 276 ], "page_no": 13 } ], "self_ref": "#/texts/138", "text": "Budiono, H. Riza, and C. Hakim (2009, August). Resource report: Building parallel text corpora for multi-domain translation system. In Proceedings of the 7th Workshop on Asian Language Resources (ALR7) , Suntec, Singapore, pp. 92-95. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Bình, V. Q. (2021). Binhvq news corpus. https://github.com/binhvq/news-corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 208.8695068359375, "coord_origin": "BOTTOMLEFT", "l": 107.56444549560547, "r": 433.4482421875, "t": 218.90380859375 }, "charspan": [ 0, 78 ], "page_no": 13 } ], "self_ref": "#/texts/139", "text": "Bình, V. Q. (2021). Binhvq news corpus. https://github.com/binhvq/news-corpus." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Carlini, N., D. Ippolito, M. Jagielski, K. Lee, F. Tramer, and C. Zhang (2022). Quantifying memorization across neural language models. arXiv preprint arXiv:2202.07646 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 180.21908569335938, "coord_origin": "BOTTOMLEFT", "l": 107.09889221191406, "r": 505.64947509765625, "t": 201.13946533203125 }, "charspan": [ 0, 169 ], "page_no": 13 } ], "self_ref": "#/texts/140", "text": "Carlini, N., D. Ippolito, M. Jagielski, K. Lee, F. Tramer, and C. Zhang (2022). Quantifying memorization across neural language models. arXiv preprint arXiv:2202.07646 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Carlini, N., C. Liu, Ú. Erlingsson, J. Kos, and D. Song (2019). The secret sharer: Evaluating and testing unintended memorization in neural networks. In 28th USENIX Security Symposium (USENIX Security 19) , pp. 267-284.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 138.99508666992188, "coord_origin": "BOTTOMLEFT", "l": 107.27818298339844, "r": 504.2413024902344, "t": 170.18450927734375 }, "charspan": [ 0, 219 ], "page_no": 13 } ], "self_ref": "#/texts/141", "text": "Carlini, N., C. Liu, Ú. Erlingsson, J. Kos, and D. Song (2019). The secret sharer: Evaluating and testing unintended memorization in neural networks. In 28th USENIX Security Symposium (USENIX Security 19) , pp. 267-284." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Carrino, C. P., C. G. Rodriguez-Penagos, and C. Armentano-Oller (2021, March). Tecla: Text classification catalan dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 109.87608337402344, "coord_origin": "BOTTOMLEFT", "l": 107.22490692138672, "r": 504.07647705078125, "t": 130.258056640625 }, "charspan": [ 0, 122 ], "page_no": 13 } ], "self_ref": "#/texts/142", "text": "Carrino, C. P., C. G. Rodriguez-Penagos, and C. Armentano-Oller (2021, March). Tecla: Text classification catalan dataset." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Caselli, T., R. Cibin, C. Conforti, E. Encinas, and M. Teli (2021). Guiding principles for participatory design-inspired natural language processing. Proceedings of the 1st Workshop on NLP for Positive Impact .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.69976806640625, "coord_origin": "BOTTOMLEFT", "l": 107.21602630615234, "r": 504.3528747558594, "t": 101.66998291015625 }, "charspan": [ 0, 210 ], "page_no": 13 } ], "self_ref": "#/texts/143", "text": "Caselli, T., R. Cibin, C. Conforti, E. Encinas, and M. Teli (2021). Guiding principles for participatory design-inspired natural language processing. Proceedings of the 1st Workshop on NLP for Positive Impact ." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "13", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.93157958984375, "r": 310.9815979003906, "t": 49.54864501953125 }, "charspan": [ 0, 2 ], "page_no": 13 } ], "self_ref": "#/texts/144", "text": "13" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "14", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.9898681640625, "r": 310.9815979003906, "t": 49.381103515625 }, "charspan": [ 0, 2 ], "page_no": 14 } ], "self_ref": "#/texts/145", "text": "14" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Gokaslan, A. and V. Cohen (2019). Openwebtext corpus. http://Skylion007.github.io/ OpenWebTextCorpus .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.4163818359375, "coord_origin": "BOTTOMLEFT", "l": 107.44552612304688, "r": 504.83184814453125, "t": 717.9892578125 }, "charspan": [ 0, 102 ], "page_no": 15 } ], "self_ref": "#/texts/146", "text": "Gokaslan, A. and V. Cohen (2019). Openwebtext corpus. http://Skylion007.github.io/ OpenWebTextCorpus ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Goldhahn, D., T. Eckart, and U. Quasthoff (2012, May). Building large monolingual dictionaries at the Leipzig corpora collection: From 100 to 200 languages. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12) , Istanbul, Turkey, pp. 759-765. European Language Resources Association (ELRA).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 644.8577880859375, "coord_origin": "BOTTOMLEFT", "l": 107.42759704589844, "r": 505.74810791015625, "t": 687.724853515625 }, "charspan": [ 0, 338 ], "page_no": 15 } ], "self_ref": "#/texts/147", "text": "Goldhahn, D., T. Eckart, and U. Quasthoff (2012, May). Building large monolingual dictionaries at the Leipzig corpora collection: From 100 to 200 languages. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12) , Istanbul, Turkey, pp. 759-765. European Language Resources Association (ELRA)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "He, W., K. Liu, J. Liu, Y. Lyu, S. Zhao, X. Xiao, Y. Liu, Y. Wang, H. Wu, Q. She, X. Liu, T. Wu, and H. Wang (2018). Dureader: a chinese machine reading comprehension dataset from real-world applications.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 603.7256469726562, "coord_origin": "BOTTOMLEFT", "l": 107.52643585205078, "r": 504.1068115234375, "t": 635.7164306640625 }, "charspan": [ 0, 204 ], "page_no": 15 } ], "self_ref": "#/texts/148", "text": "He, W., K. Liu, J. Liu, Y. Lyu, S. Zhao, X. Xiao, Y. Liu, Y. Wang, H. Wu, Q. She, X. Liu, T. Wu, and H. Wang (2018). Dureader: a chinese machine reading comprehension dataset from real-world applications." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Heafield, K. (2011, July). KenLM: Faster and smaller language model queries. In Proceedings of the Sixth Workshop on Statistical Machine Translation , Edinburgh, Scotland, pp. 187-197. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 562.6582641601562, "coord_origin": "BOTTOMLEFT", "l": 107.49415588378906, "r": 504.4220886230469, "t": 594.926513671875 }, "charspan": [ 0, 227 ], "page_no": 15 } ], "self_ref": "#/texts/149", "text": "Heafield, K. (2011, July). KenLM: Faster and smaller language model queries. In Proceedings of the Sixth Workshop on Statistical Machine Translation , Edinburgh, Scotland, pp. 187-197. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ho, V. A., D. H.-C. Nguyen, D. H. Nguyen, L. T.-V. Pham, D.-V. Nguyen, K. V. Nguyen, and N. L.-T. Nguyen (2020). Emotion recognition for vietnamese social media text.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 532.709716796875, "coord_origin": "BOTTOMLEFT", "l": 107.62950134277344, "r": 505.7392883300781, "t": 554.3221435546875 }, "charspan": [ 0, 166 ], "page_no": 15 } ], "self_ref": "#/texts/150", "text": "Ho, V. A., D. H.-C. Nguyen, D. H. Nguyen, L. T.-V. Pham, D.-V. Nguyen, K. V. Nguyen, and N. L.-T. Nguyen (2020). Emotion recognition for vietnamese social media text." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hoffmann, J., S. Borgeaud, A. Mensch, E. Buchatskaya, T. Cai, E. Rutherford, D. d. L. Casas, L. A. Hendricks, J. Welbl, A. Clark, T. Hennigan, E. Noland, K. Millican, G. v. d. Driessche, B. Damoc, A. Guy, S. Osindero, K. Simonyan, E. Elsen, J. W. Rae, O. Vinyals, and L. Sifre (2022). Training compute-optimal large language models.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 481.42333984375, "coord_origin": "BOTTOMLEFT", "l": 107.30654907226562, "r": 505.74688720703125, "t": 524.442626953125 }, "charspan": [ 0, 332 ], "page_no": 15 } ], "self_ref": "#/texts/151", "text": "Hoffmann, J., S. Borgeaud, A. Mensch, E. Buchatskaya, T. Cai, E. Rutherford, D. d. L. Casas, L. A. Hendricks, J. Welbl, A. Clark, T. Hennigan, E. Noland, K. Millican, G. v. d. Driessche, B. Damoc, A. Guy, S. Osindero, K. Simonyan, E. Elsen, J. W. Rae, O. Vinyals, and L. Sifre (2022). Training compute-optimal large language models." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Howard, J. and S. Ruder (2018). Fine-tuned language models for text classification. CoRR abs/1801.06146 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 451.5840759277344, "coord_origin": "BOTTOMLEFT", "l": 107.7571029663086, "r": 505.74505615234375, "t": 472.4336242675781 }, "charspan": [ 0, 105 ], "page_no": 15 } ], "self_ref": "#/texts/152", "text": "Howard, J. and S. Ruder (2018). Fine-tuned language models for text classification. CoRR abs/1801.06146 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jawaid, B., A. Kamran, and O. Bojar (2014). Urdu monolingual corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 431.9647521972656, "coord_origin": "BOTTOMLEFT", "l": 106.8739242553711, "r": 394.9527893066406, "t": 442.3625793457031 }, "charspan": [ 0, 68 ], "page_no": 15 } ], "self_ref": "#/texts/153", "text": "Jawaid, B., A. Kamran, and O. Bojar (2014). Urdu monolingual corpus." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jernite, Y., H. Nguyen, S. Biderman, A. Rogers, M. Masoud, V. Danchev, S. Tan, A. S. Luccioni, N. Subramani, G. Dupont, J. Dodge, K. Lo, Z. Talat, D. Radev, A. Gokaslan, S. Nikpoor, P. Henderson, R. Bommasani, and M. Mitchell (2022). Data governance in the age of large-scale data-driven language technology. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency , FAccT '22, New York, NY, USA. Association for Computing Machinery.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 369.5189208984375, "coord_origin": "BOTTOMLEFT", "l": 107.15341186523438, "r": 505.7464904785156, "t": 423.00286865234375 }, "charspan": [ 0, 465 ], "page_no": 15 } ], "self_ref": "#/texts/154", "text": "Jernite, Y., H. Nguyen, S. Biderman, A. Rogers, M. Masoud, V. Danchev, S. Tan, A. S. Luccioni, N. Subramani, G. Dupont, J. Dodge, K. Lo, Z. Talat, D. Radev, A. Gokaslan, S. Nikpoor, P. Henderson, R. Bommasani, and M. Mitchell (2022). Data governance in the age of large-scale data-driven language technology. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency , FAccT '22, New York, NY, USA. Association for Computing Machinery." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Joulin, A., E. Grave, P. Bojanowski, and T. Mikolov (2017, April). Bag of tricks for efficient text classification. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers , pp. 427-431. Association for Computational Linguistics.", "parent": { "$ref": "#/groups/2" }, "prov": [ { "bbox": { "b": 317.6898193359375, "coord_origin": "BOTTOMLEFT", "l": 107.20397186279297, "r": 504.0957946777344, "t": 360.082275390625 }, "charspan": [ 0, 308 ], "page_no": 15 } ], "self_ref": "#/texts/155", "text": "Joulin, A., E. Grave, P. Bojanowski, and T. Mikolov (2017, April). Bag of tricks for efficient text classification. In Proceedings of the 15th Conference of the European Chapter of the Association for Computational Linguistics: Volume 2, Short Papers , pp. 427-431. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kandpal, N., E. Wallace, and C. Raffel (2022). Deduplicating training data mitigates privacy risks in language models. arXiv preprint arXiv:2202.06539 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 287.60980224609375, "coord_origin": "BOTTOMLEFT", "l": 107.58759307861328, "r": 504.0298156738281, "t": 308.8553466796875 }, "charspan": [ 0, 152 ], "page_no": 15 } ], "self_ref": "#/texts/156", "text": "Kandpal, N., E. Wallace, and C. Raffel (2022). Deduplicating training data mitigates privacy risks in language models. arXiv preprint arXiv:2202.06539 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kaplan, J., S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Radford, J. Wu, and D. Amodei (2020). Scaling laws for neural language models. CoRR abs/2001.08361 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 257.923828125, "coord_origin": "BOTTOMLEFT", "l": 107.63335418701172, "r": 505.241455078125, "t": 278.91619873046875 }, "charspan": [ 0, 185 ], "page_no": 15 } ], "self_ref": "#/texts/157", "text": "Kaplan, J., S. McCandlish, T. Henighan, T. B. Brown, B. Chess, R. Child, S. Gray, A. Radford, J. Wu, and D. Amodei (2020). Scaling laws for neural language models. CoRR abs/2001.08361 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kermes, H., S. Degaetano-Ortlieb, A. Khamis, J. Knappen, and E. Teich (2016, May). The royal society corpus: From uncharted data to corpus. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16) , Portorož, Slovenia, pp. 1928-1931. European Language Resources Association (ELRA).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 206.1910858154297, "coord_origin": "BOTTOMLEFT", "l": 107.7308578491211, "r": 505.6514587402344, "t": 248.9945068359375 }, "charspan": [ 0, 324 ], "page_no": 15 } ], "self_ref": "#/texts/158", "text": "Kermes, H., S. Degaetano-Ortlieb, A. Khamis, J. Knappen, and E. Teich (2016, May). The royal society corpus: From uncharted data to corpus. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16) , Portorož, Slovenia, pp. 1928-1931. European Language Resources Association (ELRA)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kowsher, M., M. Uddin, A. Tahabilder, M. Ruhul Amin, M. F. Shahriar, and M. S. I. Sobuj (2021, September). Banglalm: Bangla corpus for language model research. Online. IEEE.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 175.93292236328125, "coord_origin": "BOTTOMLEFT", "l": 107.81340789794922, "r": 505.2470703125, "t": 197.0460205078125 }, "charspan": [ 0, 173 ], "page_no": 15 } ], "self_ref": "#/texts/159", "text": "Kowsher, M., M. Uddin, A. Tahabilder, M. Ruhul Amin, M. F. Shahriar, and M. S. I. Sobuj (2021, September). Banglalm: Bangla corpus for language model research. Online. IEEE." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kreutzer, J., I. Caswell, L. Wang, A. Wahab, D. van Esch, N. Ulzii-Orshikh, A. Tapo, N. Subramani, A. Sokolov, C. Sikasote, M. Setyawan, S. Sarin, S. Samb, B. Sagot, C. Rivera, A. Rios, I. Papadimitriou, S. Osei, P. O. Suarez, I. Orife, K. Ogueji, R. Niyongabo, T. Nguyen, M. Müller, A. Müller, S. Muhammad, N. Muhammad, A. Mnyakeni, J. Mirzakhalov, T. Matangira, C. Leong, N. Lawson, S. Kudugunta, Y. Jernite, M. Jenny, O. Firat, B. Dossou, S. Dlamini, N. de Silva, S. Çabuk Ballı, S. Biderman, A. Battisti, A. Baruwa, A. Bapna, P. Baljekar, I. Azime, A. Awokoya, D. Ataman, O. Ahia, O. Ahia, S. Agrawal, and M. Adeyemi (2022). Quality at a glance: An audit of web-crawled multilingual datasets. Transactions of the Association for Computational Linguistics 10 (0), 50-72.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.6727294921875, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.6535949707031, "t": 166.8551025390625 }, "charspan": [ 0, 773 ], "page_no": 15 } ], "self_ref": "#/texts/160", "text": "Kreutzer, J., I. Caswell, L. Wang, A. Wahab, D. van Esch, N. Ulzii-Orshikh, A. Tapo, N. Subramani, A. Sokolov, C. Sikasote, M. Setyawan, S. Sarin, S. Samb, B. Sagot, C. Rivera, A. Rios, I. Papadimitriou, S. Osei, P. O. Suarez, I. Orife, K. Ogueji, R. Niyongabo, T. Nguyen, M. Müller, A. Müller, S. Muhammad, N. Muhammad, A. Mnyakeni, J. Mirzakhalov, T. Matangira, C. Leong, N. Lawson, S. Kudugunta, Y. Jernite, M. Jenny, O. Firat, B. Dossou, S. Dlamini, N. de Silva, S. Çabuk Ballı, S. Biderman, A. Battisti, A. Baruwa, A. Bapna, P. Baljekar, I. Azime, A. Awokoya, D. Ataman, O. Ahia, O. Ahia, S. Agrawal, and M. Adeyemi (2022). Quality at a glance: An audit of web-crawled multilingual datasets. Transactions of the Association for Computational Linguistics 10 (0), 50-72." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "15", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.9464416503906, "r": 310.9815673828125, "t": 49.6328125 }, "charspan": [ 0, 2 ], "page_no": 15 } ], "self_ref": "#/texts/161", "text": "15" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kudo, T. (2018, July). Subword regularization: Improving neural network translation models with multiple subword candidates. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , Melbourne, Australia, pp. 66-75. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 674.5964965820312, "coord_origin": "BOTTOMLEFT", "l": 107.65170288085938, "r": 504.4319763183594, "t": 717.5701293945312 }, "charspan": [ 0, 317 ], "page_no": 16 } ], "self_ref": "#/texts/162", "text": "Kudo, T. (2018, July). Subword regularization: Improving neural network translation models with multiple subword candidates. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , Melbourne, Australia, pp. 66-75. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kunchukuttan, A. (2020). The IndicNLP Library. https://github.com/anoopkunchukuttan/indic_ nlp_library/blob/master/docs/indicnlp.pdf .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 644.78759765625, "coord_origin": "BOTTOMLEFT", "l": 107.74006652832031, "r": 504.5616455078125, "t": 665.6636962890625 }, "charspan": [ 0, 134 ], "page_no": 16 } ], "self_ref": "#/texts/163", "text": "Kunchukuttan, A. (2020). The IndicNLP Library. https://github.com/anoopkunchukuttan/indic_ nlp_library/blob/master/docs/indicnlp.pdf ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kunchukuttan, A., D. Kakwani, S. Golla, G. N.C., A. Bhattacharyya, M. M. Khapra, and P. Kumar (2020). Ai4bharat-indicnlp corpus: Monolingual corpora and word embeddings for indic languages. arXiv preprint arXiv:2005.00085 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 603.8220825195312, "coord_origin": "BOTTOMLEFT", "l": 107.69255828857422, "r": 505.7466735839844, "t": 635.5896606445312 }, "charspan": [ 0, 223 ], "page_no": 16 } ], "self_ref": "#/texts/164", "text": "Kunchukuttan, A., D. Kakwani, S. Golla, G. N.C., A. Bhattacharyya, M. M. Khapra, and P. Kumar (2020). Ai4bharat-indicnlp corpus: Monolingual corpora and word embeddings for indic languages. arXiv preprint arXiv:2005.00085 ." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Kunchukuttan, A., P. Mehta, and P. Bhattacharyya (2018, May). The IIT Bombay English-Hindi parallel corpus. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018) , Miyazaki, Japan. European Language Resources Association (ELRA).", "parent": { "$ref": "#/groups/3" }, "prov": [ { "bbox": { "b": 551.9140625, "coord_origin": "BOTTOMLEFT", "l": 107.64632415771484, "r": 504.0817565917969, "t": 594.16943359375 }, "charspan": [ 0, 279 ], "page_no": 16 } ], "self_ref": "#/texts/165", "text": "Kunchukuttan, A., P. Mehta, and P. Bhattacharyya (2018, May). The IIT Bombay English-Hindi parallel corpus. In Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018) , Miyazaki, Japan. European Language Resources Association (ELRA)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kurniawan, K. and S. Louvan (2018). Indosum: A new benchmark dataset for indonesian text summarization. In 2018 International Conference on Asian Language Processing (IALP) , pp. 215-220. IEEE.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 510.83306884765625, "coord_origin": "BOTTOMLEFT", "l": 107.59474182128906, "r": 505.7701721191406, "t": 542.0592651367188 }, "charspan": [ 0, 193 ], "page_no": 16 } ], "self_ref": "#/texts/166", "text": "Kurniawan, K. and S. Louvan (2018). Indosum: A new benchmark dataset for indonesian text summarization. In 2018 International Conference on Asian Language Processing (IALP) , pp. 215-220. IEEE." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Külebi, B. (2021, October). ParlamentParla - Speech corpus of Catalan Parliamentary sessions.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 491.570068359375, "coord_origin": "BOTTOMLEFT", "l": 107.5748062133789, "r": 487.7145690917969, "t": 501.4150695800781 }, "charspan": [ 0, 93 ], "page_no": 16 } ], "self_ref": "#/texts/167", "text": "Külebi, B. (2021, October). ParlamentParla - Speech corpus of Catalan Parliamentary sessions." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Leahy, C. and S. Biderman (2021). The hard problem of aligning AI to human values. In The State of AI Ethics Report , Volume 4, pp. 180-183. The Montreal AI Ethics Institute.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 460.9255065917969, "coord_origin": "BOTTOMLEFT", "l": 107.50896453857422, "r": 505.6374816894531, "t": 482.26177978515625 }, "charspan": [ 0, 174 ], "page_no": 16 } ], "self_ref": "#/texts/168", "text": "Leahy, C. and S. Biderman (2021). The hard problem of aligning AI to human values. In The State of AI Ethics Report , Volume 4, pp. 180-183. The Montreal AI Ethics Institute." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Lee, K., D. Ippolito, A. Nystrom, C. Zhang, D. Eck, C. Callison-Burch, and N. Carlini (2022). Deduplicating training data makes language models better. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics . Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 420.23724365234375, "coord_origin": "BOTTOMLEFT", "l": 107.50053405761719, "r": 505.6509094238281, "t": 451.6004943847656 }, "charspan": [ 0, 287 ], "page_no": 16 } ], "self_ref": "#/texts/169", "text": "Lee, K., D. Ippolito, A. Nystrom, C. Zhang, D. Eck, C. Callison-Burch, and N. Carlini (2022). Deduplicating training data makes language models better. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics . Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Lhoest, Q., A. Villanova del Moral, Y. Jernite, A. Thakur, P. von Platen, S. Patil, J. Chaumond, M. Drame, J. Plu, L. Tunstall, J. Davison, M. Šaško, G. Chhablani, B. Malik, S. Brandeis, T. Le Scao, V. Sanh, C. Xu, N. Patry, A. McMillan-Major, P. Schmid, S. Gugger, C. Delangue, T. Matussière, L. Debut, S. Bekman, P. Cistac, T. Goehringer, V. Mustar, F. Lagunas, A. Rush, and T. Wolf (2021, November). Datasets: A community library for natural language processing. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations , Online and Punta Cana, Dominican Republic, pp. 175-184. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 324.2727966308594, "coord_origin": "BOTTOMLEFT", "l": 107.25635528564453, "r": 505.74127197265625, "t": 410.7928161621094 }, "charspan": [ 0, 679 ], "page_no": 16 } ], "self_ref": "#/texts/170", "text": "Lhoest, Q., A. Villanova del Moral, Y. Jernite, A. Thakur, P. von Platen, S. Patil, J. Chaumond, M. Drame, J. Plu, L. Tunstall, J. Davison, M. Šaško, G. Chhablani, B. Malik, S. Brandeis, T. Le Scao, V. Sanh, C. Xu, N. Patry, A. McMillan-Major, P. Schmid, S. Gugger, C. Delangue, T. Matussière, L. Debut, S. Bekman, P. Cistac, T. Goehringer, V. Mustar, F. Lagunas, A. Rush, and T. Wolf (2021, November). Datasets: A community library for natural language processing. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations , Online and Punta Cana, Dominican Republic, pp. 175-184. Association for Computational Linguistics." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Li, Y., D. Choi, J. Chung, N. Kushman, J. Schrittwieser, R. Leblond, T. Eccles, J. Keeling, F. Gimeno, A. D. Lago, T. Hubert, P. Choy, C. d. M. d'Autume, I. Babuschkin, X. Chen, P.-S. Huang, J. Welbl, S. Gowal, A. Cherepanov, J. Molloy, D. J. Mankowitz, E. S. Robson, P. Kohli, N. de Freitas, K. Kavukcuoglu, and O. Vinyals (2022). Competition-level code generation with alphacode.", "parent": { "$ref": "#/groups/4" }, "prov": [ { "bbox": { "b": 272.17034912109375, "coord_origin": "BOTTOMLEFT", "l": 107.15587615966797, "r": 505.6770935058594, "t": 314.74359130859375 }, "charspan": [ 0, 381 ], "page_no": 16 } ], "self_ref": "#/texts/171", "text": "Li, Y., D. Choi, J. Chung, N. Kushman, J. Schrittwieser, R. Leblond, T. Eccles, J. Keeling, F. Gimeno, A. D. Lago, T. Hubert, P. Choy, C. d. M. d'Autume, I. Babuschkin, X. Chen, P.-S. Huang, J. Welbl, S. Gowal, A. Cherepanov, J. Molloy, D. J. Mankowitz, E. S. Robson, P. Kohli, N. de Freitas, K. Kavukcuoglu, and O. Vinyals (2022). Competition-level code generation with alphacode." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Lieber, O., O. Sharir, B. Lenz, and Y. Shoham (2021). Jurassic-1: Technical details and evaluation. White Paper. AI21 Labs .", "parent": { "$ref": "#/groups/4" }, "prov": [ { "bbox": { "b": 242.07720947265625, "coord_origin": "BOTTOMLEFT", "l": 107.5421371459961, "r": 505.7395935058594, "t": 262.8277587890625 }, "charspan": [ 0, 124 ], "page_no": 16 } ], "self_ref": "#/texts/172", "text": "Lieber, O., O. Sharir, B. Lenz, and Y. Shoham (2021). Jurassic-1: Technical details and evaluation. White Paper. AI21 Labs ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Lison, P. and J. Tiedemann (2016). Opensubtitles2016: Extracting large parallel corpora from movie and tv subtitles.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 212.35508728027344, "coord_origin": "BOTTOMLEFT", "l": 107.54146575927734, "r": 504.03656005859375, "t": 232.78955078125 }, "charspan": [ 0, 116 ], "page_no": 16 } ], "self_ref": "#/texts/173", "text": "Lison, P. and J. Tiedemann (2016). Opensubtitles2016: Extracting large parallel corpora from movie and tv subtitles." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Lo, K., L. L. Wang, M. Neumann, R. Kinney, and D. Weld (2020, July). S2ORC: The semantic scholar open research corpus. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics , Online, pp. 4969-4983. Association for Computational Linguistics.", "parent": { "$ref": "#/groups/5" }, "prov": [ { "bbox": { "b": 170.88812255859375, "coord_origin": "BOTTOMLEFT", "l": 107.50669860839844, "r": 504.4978942871094, "t": 202.85504150390625 }, "charspan": [ 0, 277 ], "page_no": 16 } ], "self_ref": "#/texts/174", "text": "Lo, K., L. L. Wang, M. Neumann, R. Kinney, and D. Weld (2020, July). S2ORC: The semantic scholar open research corpus. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics , Online, pp. 4969-4983. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Lopes, C. V., P. Maj, P. Martins, V. Saini, D. Yang, J. Zitny, H. Sajnani, and J. Vitek (2017). Déjàvu: a map of code duplicates on github. Proceedings of the ACM on Programming Languages 1 (OOPSLA), 1-28.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 130.19207763671875, "coord_origin": "BOTTOMLEFT", "l": 107.65580749511719, "r": 505.74505615234375, "t": 161.56884765625 }, "charspan": [ 0, 205 ], "page_no": 16 } ], "self_ref": "#/texts/175", "text": "Lopes, C. V., P. Maj, P. Martins, V. Saini, D. Yang, J. Zitny, H. Sajnani, and J. Vitek (2017). Déjàvu: a map of code duplicates on github. Proceedings of the ACM on Programming Languages 1 (OOPSLA), 1-28." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Luccioni, A. S. and J. D. Viviano (2021). What's in the box? a preliminary analysis of undesirable content in the common crawl corpus. Published in the Proceedings of ACL 2021 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 99.64605712890625, "coord_origin": "BOTTOMLEFT", "l": 107.46929168701172, "r": 503.9989929199219, "t": 120.93511962890625 }, "charspan": [ 0, 177 ], "page_no": 16 } ], "self_ref": "#/texts/176", "text": "Luccioni, A. S. and J. D. Viviano (2021). What's in the box? a preliminary analysis of undesirable content in the common crawl corpus. Published in the Proceedings of ACL 2021 ." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Mahendra, R., A. F. Aji, S. Louvan, F. Rahman, and C. Vania (2021). Indonli: A natural language inference dataset for indonesian. arXiv preprint arXiv:2110.14566 .", "parent": { "$ref": "#/groups/6" }, "prov": [ { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 107.2805404663086, "r": 503.99468994140625, "t": 90.1417236328125 }, "charspan": [ 0, 163 ], "page_no": 16 } ], "self_ref": "#/texts/177", "text": "Mahendra, R., A. F. Aji, S. Louvan, F. Rahman, and C. Vania (2021). Indonli: A natural language inference dataset for indonesian. arXiv preprint arXiv:2110.14566 ." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "16", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 301.0106201171875, "r": 311.2333679199219, "t": 49.58758544921875 }, "charspan": [ 0, 2 ], "page_no": 16 } ], "self_ref": "#/texts/178", "text": "16" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Manber, U. and G. Myers (1993). Suffix arrays: A new method for on-line string searches. SIAM Journal on Computing 22 (5), 935-948.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.97607421875, "coord_origin": "BOTTOMLEFT", "l": 107.26850128173828, "r": 504.5064392089844, "t": 717.6826782226562 }, "charspan": [ 0, 131 ], "page_no": 17 } ], "self_ref": "#/texts/179", "text": "Manber, U. and G. Myers (1993). Suffix arrays: A new method for on-line string searches. SIAM Journal on Computing 22 (5), 935-948." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Manku, G. S., A. Jain, and A. Das Sarma (2007). Detecting near-duplicates for web crawling. In Proceedings of the 16th International Conference on World Wide Web , WWW '07, New York, NY, USA, pp. 141-150. Association for Computing Machinery.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 654.8859252929688, "coord_origin": "BOTTOMLEFT", "l": 107.26089477539062, "r": 505.2431945800781, "t": 686.9276733398438 }, "charspan": [ 0, 241 ], "page_no": 17 } ], "self_ref": "#/texts/180", "text": "Manku, G. S., A. Jain, and A. Das Sarma (2007). Detecting near-duplicates for web crawling. In Proceedings of the 16th International Conference on World Wide Web , WWW '07, New York, NY, USA, pp. 141-150. Association for Computing Machinery." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Mayeesha, T. T., A. M. Sarwar, and R. M. Rahman (2020, November). Deep learning based question answering system in Bengali.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 623.9600830078125, "coord_origin": "BOTTOMLEFT", "l": 107.14989471435547, "r": 504.0306091308594, "t": 644.9791259765625 }, "charspan": [ 0, 123 ], "page_no": 17 } ], "self_ref": "#/texts/181", "text": "Mayeesha, T. T., A. M. Sarwar, and R. M. Rahman (2020, November). Deep learning based question answering system in Bengali." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "McMillan-Major, A., Z. Alyafeai, S. Biderman, K. Chen, F. De Toni, G. Dupont, H. Elsahar, C. Emezue, A. F. Aji, S. Ili'c, N. Khamis, C. Leong, M. Masoud, A. Soroa, P. O. Suarez, Z. Talat, D. van Strien, and Y. Jernite (2022). Documenting geographically and contextually diverse data sources: The bigscience catalogue of language data and resources.", "parent": { "$ref": "#/groups/7" }, "prov": [ { "bbox": { "b": 571.6458129882812, "coord_origin": "BOTTOMLEFT", "l": 107.18138885498047, "r": 505.2942810058594, "t": 614.2986450195312 }, "charspan": [ 0, 348 ], "page_no": 17 } ], "self_ref": "#/texts/182", "text": "McMillan-Major, A., Z. Alyafeai, S. Biderman, K. Chen, F. De Toni, G. Dupont, H. Elsahar, C. Emezue, A. F. Aji, S. Ili'c, N. Khamis, C. Leong, M. Masoud, A. Soroa, P. O. Suarez, Z. Talat, D. van Strien, and Y. Jernite (2022). Documenting geographically and contextually diverse data sources: The bigscience catalogue of language data and resources." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Moeljadi, D. (2012). Usage of indonesian possessive verbal predicates: A statistical analysis based on questionnaire and storytelling surveys. In 5th Conference on Austronesian and Papuan Languages and Linguistics (APLL5) , SOAS, University of London.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 530.2640991210938, "coord_origin": "BOTTOMLEFT", "l": 107.28323364257812, "r": 504.0014343261719, "t": 561.6052856445312 }, "charspan": [ 0, 251 ], "page_no": 17 } ], "self_ref": "#/texts/183", "text": "Moeljadi, D. (2012). Usage of indonesian possessive verbal predicates: A statistical analysis based on questionnaire and storytelling surveys. In 5th Conference on Austronesian and Papuan Languages and Linguistics (APLL5) , SOAS, University of London." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Mohr, G., J. Kunze, and M. Stack (2008). The warc file format 1.0 (iso 28500).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 510.40411376953125, "coord_origin": "BOTTOMLEFT", "l": 107.13787078857422, "r": 424.1531982421875, "t": 520.32568359375 }, "charspan": [ 0, 78 ], "page_no": 17 } ], "self_ref": "#/texts/184", "text": "Mohr, G., J. Kunze, and M. Stack (2008). The warc file format 1.0 (iso 28500)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Nekoto, W., V. Marivate, T. Matsila, T. E. Fasubaa, T. Fagbohungbe, S. O. Akinola, S. H. Muhammad, S. K. Kabenamualu, S. Osei, F. Sackey, R. A. Niyongabo, R. Macharm, P. Ogayo, O. Ahia, M. M. Berhe, M. Adeyemi, M. Mokgesi-Selinga, L. Okegbemi, L. Martinus, K. Tajudeen, K. Degila, K. Ogueji, K. Siminyu, J. Kreutzer, J. Webster, J. T. Ali, J. Z. Abbott, I. Orife, I. Ezeani, I. A. Dangana, H. Kamper, H. Elsahar, G. Duru, G. Kioko, E. Murhabazi, E. V. Biljon, D. Whitenack, C. Onyefuluchi, C. C. Emezue, B. F. P. Dossou, B. Sibanda, B. I. Bassey, A. Olabiyi, A. Ramkilowan, A. Öktem, A. Akinfaderin, and A. Bashir (2020). Participatory research for low-resourced machine translation: A case study in african languages. In T. Cohn, Y. He, and Y. Liu (Eds.), Findings of the Association for Computational Linguistics: EMNLP 2020, Online Event, 16-20 November 2020 , Volume EMNLP 2020 of Findings of ACL , pp. 2144-2160. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 380.8861389160156, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.7987365722656, "t": 500.0885925292969 }, "charspan": [ 0, 960 ], "page_no": 17 } ], "self_ref": "#/texts/185", "text": "Nekoto, W., V. Marivate, T. Matsila, T. E. Fasubaa, T. Fagbohungbe, S. O. Akinola, S. H. Muhammad, S. K. Kabenamualu, S. Osei, F. Sackey, R. A. Niyongabo, R. Macharm, P. Ogayo, O. Ahia, M. M. Berhe, M. Adeyemi, M. Mokgesi-Selinga, L. Okegbemi, L. Martinus, K. Tajudeen, K. Degila, K. Ogueji, K. Siminyu, J. Kreutzer, J. Webster, J. T. Ali, J. Z. Abbott, I. Orife, I. Ezeani, I. A. Dangana, H. Kamper, H. Elsahar, G. Duru, G. Kioko, E. Murhabazi, E. V. Biljon, D. Whitenack, C. Onyefuluchi, C. C. Emezue, B. F. P. Dossou, B. Sibanda, B. I. Bassey, A. Olabiyi, A. Ramkilowan, A. Öktem, A. Akinfaderin, and A. Bashir (2020). Participatory research for low-resourced machine translation: A case study in african languages. In T. Cohn, Y. He, and Y. Liu (Eds.), Findings of the Association for Computational Linguistics: EMNLP 2020, Online Event, 16-20 November 2020 , Volume EMNLP 2020 of Findings of ACL , pp. 2144-2160. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ngo, C. and T. H. Trinh (2021). Styled augmented translation (sat). https://github.com/vietai/SAT .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 361.41265869140625, "coord_origin": "BOTTOMLEFT", "l": 107.16851043701172, "r": 500.11566162109375, "t": 371.63446044921875 }, "charspan": [ 0, 99 ], "page_no": 17 } ], "self_ref": "#/texts/186", "text": "Ngo, C. and T. H. Trinh (2021). Styled augmented translation (sat). https://github.com/vietai/SAT ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Nguyen, K. V., V. D. Nguyen, P. X. V. Nguyen, T. T. H. Truong, and N. L.-T. Nguyen (2018). Uit-vsfc: Vietnamese students' feedback corpus for sentiment analysis. In 2018 10th International Conference on Knowledge and Systems Engineering (KSE) , pp. 19-24.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 319.3028564453125, "coord_origin": "BOTTOMLEFT", "l": 107.33122253417969, "r": 505.7449645996094, "t": 351.4919738769531 }, "charspan": [ 0, 255 ], "page_no": 17 } ], "self_ref": "#/texts/187", "text": "Nguyen, K. V., V. D. Nguyen, P. X. V. Nguyen, T. T. H. Truong, and N. L.-T. Nguyen (2018). Uit-vsfc: Vietnamese students' feedback corpus for sentiment analysis. In 2018 10th International Conference on Knowledge and Systems Engineering (KSE) , pp. 19-24." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Nguyen, T., H. Pham, M. Truong, H. Duc, and P. Tan (2021). Vietnamese poem generator. https://github.com/fsoft-ailab/Poem-Generator.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 288.8136291503906, "coord_origin": "BOTTOMLEFT", "l": 107.57347106933594, "r": 505.7450256347656, "t": 309.7726135253906 }, "charspan": [ 0, 132 ], "page_no": 17 } ], "self_ref": "#/texts/188", "text": "Nguyen, T., H. Pham, M. Truong, H. Duc, and P. Tan (2021). Vietnamese poem generator. https://github.com/fsoft-ailab/Poem-Generator." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Nomoto, H., K. Okano, D. Moeljadi, and H. Sawada (2018). Tufs asian language parallel corpus (talpco). In Proceedings of the Twenty-Fourth Annual Meeting of the Association for Natural Language Processing , pp. 436-439. Association for Natural Language Processing.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 246.73974609375, "coord_origin": "BOTTOMLEFT", "l": 107.4731216430664, "r": 504.3941650390625, "t": 278.88372802734375 }, "charspan": [ 0, 264 ], "page_no": 17 } ], "self_ref": "#/texts/189", "text": "Nomoto, H., K. Okano, D. Moeljadi, and H. Sawada (2018). Tufs asian language parallel corpus (talpco). In Proceedings of the Twenty-Fourth Annual Meeting of the Association for Natural Language Processing , pp. 436-439. Association for Natural Language Processing." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ortiz Suárez, P. J., L. Romary, and B. Sagot (2020, July). A monolingual approach to contextualized word embeddings for mid-resource languages. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics , Online, pp. 1703-1714. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 194.882080078125, "coord_origin": "BOTTOMLEFT", "l": 107.5305404663086, "r": 504.1767578125, "t": 237.52667236328125 }, "charspan": [ 0, 302 ], "page_no": 17 } ], "self_ref": "#/texts/190", "text": "Ortiz Suárez, P. J., L. Romary, and B. Sagot (2020, July). A monolingual approach to contextualized word embeddings for mid-resource languages. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics , Online, pp. 1703-1714. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ortiz Suárez, P. J., B. Sagot, and L. Romary (2019). Asynchronous pipelines for processing huge corpora on medium to low resource infrastructures. Proceedings of the Workshop on Challenges in the Management of Large Corpora (CMLC-7) 2019. Cardiff, 22nd July 2019, Mannheim, pp. 9 16. Leibniz-Institut für Deutsche Sprache.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 142.2950897216797, "coord_origin": "BOTTOMLEFT", "l": 107.39585876464844, "r": 505.4946594238281, "t": 184.70819091796875 }, "charspan": [ 0, 322 ], "page_no": 17 } ], "self_ref": "#/texts/191", "text": "Ortiz Suárez, P. J., B. Sagot, and L. Romary (2019). Asynchronous pipelines for processing huge corpora on medium to low resource infrastructures. Proceedings of the Workshop on Challenges in the Management of Large Corpora (CMLC-7) 2019. Cardiff, 22nd July 2019, Mannheim, pp. 9 16. Leibniz-Institut für Deutsche Sprache." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Parida, S., O. Bojar, and S. R. Dash (2020). Odiencorp: Odia-english and odia-only corpus for machine translation. In Smart Intelligent Computing and Applications , pp. 495-504. Springer.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 111.499267578125, "coord_origin": "BOTTOMLEFT", "l": 107.58297729492188, "r": 504.9048767089844, "t": 132.2037353515625 }, "charspan": [ 0, 187 ], "page_no": 17 } ], "self_ref": "#/texts/192", "text": "Parida, S., O. Bojar, and S. R. Dash (2020). Odiencorp: Odia-english and odia-only corpus for machine translation. In Smart Intelligent Computing and Applications , pp. 495-504. Springer." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Pisceldo, F., R. Manurung, and M. Adriani (2009). Probabilistic part-of-speech tagging for bahasa indonesia. In Third International Workshop on Malay and Indonesian Language Engineering (MALINDO) , Suntec, Singapore.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.77777099609375, "coord_origin": "BOTTOMLEFT", "l": 107.8388900756836, "r": 504.2226257324219, "t": 101.544677734375 }, "charspan": [ 0, 216 ], "page_no": 17 } ], "self_ref": "#/texts/193", "text": "Pisceldo, F., R. Manurung, and M. Adriani (2009). Probabilistic part-of-speech tagging for bahasa indonesia. In Third International Workshop on Malay and Indonesian Language Engineering (MALINDO) , Suntec, Singapore." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "17", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.9245300292969, "r": 310.9815673828125, "t": 49.533447265625 }, "charspan": [ 0, 2 ], "page_no": 17 } ], "self_ref": "#/texts/194", "text": "17" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Qi, P., Y. Zhang, Y. Zhang, J. Bolton, and C. D. Manning (2020). Stanza: A Python natural language processing toolkit for many human languages. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations .", "parent": { "$ref": "#/groups/8" }, "prov": [ { "bbox": { "b": 685.8038330078125, "coord_origin": "BOTTOMLEFT", "l": 107.4608154296875, "r": 504.089599609375, "t": 717.62158203125 }, "charspan": [ 0, 259 ], "page_no": 18 } ], "self_ref": "#/texts/195", "text": "Qi, P., Y. Zhang, Y. Zhang, J. Bolton, and C. D. Manning (2020). Stanza: A Python natural language processing toolkit for many human languages. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Radford, A., J. Wu, R. Child, D. Luan, D. Amodei, and I. Sutskever (2019). Language models are unsupervised multitask learners.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 656.8570556640625, "coord_origin": "BOTTOMLEFT", "l": 107.55138397216797, "r": 503.9975891113281, "t": 677.6571044921875 }, "charspan": [ 0, 127 ], "page_no": 18 } ], "self_ref": "#/texts/196", "text": "Radford, A., J. Wu, R. Child, D. Luan, D. Amodei, and I. Sutskever (2019). Language models are unsupervised multitask learners." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Rae, J. W., S. Borgeaud, T. Cai, K. Millican, J. Hoffmann, H. F. Song, J. Aslanides, S. Henderson, R. Ring, S. Young, E. Rutherford, T. Hennigan, J. Menick, A. Cassirer, R. Powell, G. van den Driessche, L. A. Hendricks, M. Rauh, P. Huang, A. Glaese, J. Welbl, S. Dathathri, S. Huang, J. Uesato, J. Mellor, I. Higgins, A. Creswell, N. McAleese, A. Wu, E. Elsen, S. M. Jayakumar, E. Buchatskaya, D. Budden, E. Sutherland, K. Simonyan, M. Paganini, L. Sifre, L. Martens, X. L. Li, A. Kuncoro, A. Nematzadeh, E. Gribovskaya, D. Donato, A. Lazaridou, A. Mensch, J. Lespiau, M. Tsimpoukelli, N. Grigorev, D. Fritz, T. Sottiaux, M. Pajarskas, T. Pohlen, Z. Gong, D. Toyama, C. de Masson d'Autume, Y. Li, T. Terzi, V. Mikulik, I. Babuschkin, A. Clark, D. de Las Casas, A. Guy, C. Jones, J. Bradbury, M. Johnson, B. A. Hechtman, L. Weidinger, I. Gabriel, W. S. Isaac, E. Lockhart, S. Osindero, L. Rimell, C. Dyer, O. Vinyals, K. Ayoub, J. Stanway, L. Bennett, D. Hassabis, K. Kavukcuoglu, and G. Irving (2021). Scaling language models: Methods, analysis & insights from training gopher. CoRR abs/2112.11446 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 518.2421875, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74346923828125, "t": 648.0121459960938 }, "charspan": [ 0, 1099 ], "page_no": 18 } ], "self_ref": "#/texts/197", "text": "Rae, J. W., S. Borgeaud, T. Cai, K. Millican, J. Hoffmann, H. F. Song, J. Aslanides, S. Henderson, R. Ring, S. Young, E. Rutherford, T. Hennigan, J. Menick, A. Cassirer, R. Powell, G. van den Driessche, L. A. Hendricks, M. Rauh, P. Huang, A. Glaese, J. Welbl, S. Dathathri, S. Huang, J. Uesato, J. Mellor, I. Higgins, A. Creswell, N. McAleese, A. Wu, E. Elsen, S. M. Jayakumar, E. Buchatskaya, D. Budden, E. Sutherland, K. Simonyan, M. Paganini, L. Sifre, L. Martens, X. L. Li, A. Kuncoro, A. Nematzadeh, E. Gribovskaya, D. Donato, A. Lazaridou, A. Mensch, J. Lespiau, M. Tsimpoukelli, N. Grigorev, D. Fritz, T. Sottiaux, M. Pajarskas, T. Pohlen, Z. Gong, D. Toyama, C. de Masson d'Autume, Y. Li, T. Terzi, V. Mikulik, I. Babuschkin, A. Clark, D. de Las Casas, A. Guy, C. Jones, J. Bradbury, M. Johnson, B. A. Hechtman, L. Weidinger, I. Gabriel, W. S. Isaac, E. Lockhart, S. Osindero, L. Rimell, C. Dyer, O. Vinyals, K. Ayoub, J. Stanway, L. Bennett, D. Hassabis, K. Kavukcuoglu, and G. Irving (2021). Scaling language models: Methods, analysis & insights from training gopher. CoRR abs/2112.11446 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Raffel, C., N. Shazeer, A. Roberts, K. Lee, S. Narang, M. Matena, Y. Zhou, W. Li, and P. J. Liu (2020). Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learning Research 21 , 1-67.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 478.2991943359375, "coord_origin": "BOTTOMLEFT", "l": 107.67035675048828, "r": 505.1582336425781, "t": 510.3704528808594 }, "charspan": [ 0, 234 ], "page_no": 18 } ], "self_ref": "#/texts/198", "text": "Raffel, C., N. Shazeer, A. Roberts, K. Lee, S. Narang, M. Matena, Y. Zhou, W. Li, and P. J. Liu (2020). Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learning Research 21 , 1-67." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Rahman, M., E. Kumar Dey, et al. (2018). Datasets for aspect-based sentiment analysis in bangla and its baseline evaluation. Data 3 (2), 15.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 449.22509765625, "coord_origin": "BOTTOMLEFT", "l": 107.50702667236328, "r": 504.00543212890625, "t": 469.5616149902344 }, "charspan": [ 0, 140 ], "page_no": 18 } ], "self_ref": "#/texts/199", "text": "Rahman, M., E. Kumar Dey, et al. (2018). Datasets for aspect-based sentiment analysis in bangla and its baseline evaluation. Data 3 (2), 15." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Rahutomo, F. and A. Miqdad Muadz Muzad (2018). Indonesian news corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 430.8983154296875, "coord_origin": "BOTTOMLEFT", "l": 107.27254486083984, "r": 417.6774597167969, "t": 440.7951354980469 }, "charspan": [ 0, 70 ], "page_no": 18 } ], "self_ref": "#/texts/200", "text": "Rahutomo, F. and A. Miqdad Muadz Muzad (2018). Indonesian news corpus." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Rajkomar, A., M. Hardt, M. D. Howell, G. Corrado, and M. H. Chin (2018). Ensuring fairness in machine learning to advance health equity. Annals of Internal Medicine 169 , 866-872.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 401.7140808105469, "coord_origin": "BOTTOMLEFT", "l": 107.68900299072266, "r": 503.99725341796875, "t": 422.1111755371094 }, "charspan": [ 0, 179 ], "page_no": 18 } ], "self_ref": "#/texts/201", "text": "Rajkomar, A., M. Hardt, M. D. Howell, G. Corrado, and M. H. Chin (2018). Ensuring fairness in machine learning to advance health equity. Annals of Internal Medicine 169 , 866-872." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ramesh, G., S. Doddapaneni, A. Bheemaraj, M. Jobanputra, R. AK, A. Sharma, S. Sahoo, H. Diddee, M. J, D. Kakwani, N. Kumar, A. Pradeep, S. Nagaraj, K. Deepak, V. Raghavan, A. Kunchukuttan, P. Kumar, and M. S. Khapra (2021). Samanantar: The largest publicly available parallel corpora collection for 11 indic languages.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 350.3567199707031, "coord_origin": "BOTTOMLEFT", "l": 107.88495635986328, "r": 505.24969482421875, "t": 392.71826171875 }, "charspan": [ 0, 318 ], "page_no": 18 } ], "self_ref": "#/texts/202", "text": "Ramesh, G., S. Doddapaneni, A. Bheemaraj, M. Jobanputra, R. AK, A. Sharma, S. Sahoo, H. Diddee, M. J, D. Kakwani, N. Kumar, A. Pradeep, S. Nagaraj, K. Deepak, V. Raghavan, A. Kunchukuttan, P. Kumar, and M. S. Khapra (2021). Samanantar: The largest publicly available parallel corpora collection for 11 indic languages." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Rodriguez-Penagos, C. G. and C. Armentano-Oller (2021a, June). Enriched conllu ancora for ml training.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 320.88116455078125, "coord_origin": "BOTTOMLEFT", "l": 107.86164855957031, "r": 503.99725341796875, "t": 342.2257385253906 }, "charspan": [ 0, 102 ], "page_no": 18 } ], "self_ref": "#/texts/203", "text": "Rodriguez-Penagos, C. G. and C. Armentano-Oller (2021a, June). Enriched conllu ancora for ml training." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Rodriguez-Penagos, C. G. and C. Armentano-Oller (2021b, February). VilaQuAD: an extractive QA dataset from Catalan newswire.", "parent": { "$ref": "#/groups/9" }, "prov": [ { "bbox": { "b": 292.2640686035156, "coord_origin": "BOTTOMLEFT", "l": 107.7324447631836, "r": 504.4914855957031, "t": 312.7929992675781 }, "charspan": [ 0, 124 ], "page_no": 18 } ], "self_ref": "#/texts/204", "text": "Rodriguez-Penagos, C. G. and C. Armentano-Oller (2021b, February). VilaQuAD: an extractive QA dataset from Catalan newswire." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Rodriguez-Penagos, C. G. and C. Armentano-Oller (2021c, February). ViquiQuAD: an extractive QA dataset from Catalan Wikipedia.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 262.92144775390625, "coord_origin": "BOTTOMLEFT", "l": 107.78450012207031, "r": 503.9947814941406, "t": 283.4704284667969 }, "charspan": [ 0, 126 ], "page_no": 18 } ], "self_ref": "#/texts/205", "text": "Rodriguez-Penagos, C. G. and C. Armentano-Oller (2021c, February). ViquiQuAD: an extractive QA dataset from Catalan Wikipedia." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Rogers, A. (2021, August). Changing the world by changing the data. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) , Online, pp. 2182-2194. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 211.50244140625, "coord_origin": "BOTTOMLEFT", "l": 107.90819549560547, "r": 505.7440490722656, "t": 254.1778564453125 }, "charspan": [ 0, 325 ], "page_no": 18 } ], "self_ref": "#/texts/206", "text": "Rogers, A. (2021, August). Changing the world by changing the data. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) , Online, pp. 2182-2194. Association for Computational Linguistics." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Sajjad, H., A. Abdelali, N. Durrani, and F. Dalvi (2020, December). AraBench: Benchmarking dialectal Arabic-English machine translation. In Proceedings of the 28th International Conference on Computational Linguistics , Barcelona, Spain (Online), pp. 5094-5107. International Committee on Computational Linguistics.", "parent": { "$ref": "#/groups/10" }, "prov": [ { "bbox": { "b": 160.99607849121094, "coord_origin": "BOTTOMLEFT", "l": 107.48924255371094, "r": 504.0653991699219, "t": 203.446533203125 }, "charspan": [ 0, 315 ], "page_no": 18 } ], "self_ref": "#/texts/207", "text": "Sajjad, H., A. Abdelali, N. Durrani, and F. Dalvi (2020, December). AraBench: Benchmarking dialectal Arabic-English machine translation. In Proceedings of the 28th International Conference on Computational Linguistics , Barcelona, Spain (Online), pp. 5094-5107. International Committee on Computational Linguistics." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "Sakti, S., E. Kelana, H. Riza, S. Sakai, K. Markov, and S. Nakamura (2008). Development of Indonesian large vocabulary continuous speech recognition system within a-STAR project. In Proceedings of the Workshop on Technologies and Corpora for Asia-Pacific Speech Translation (TCAST) .", "parent": { "$ref": "#/groups/10" }, "prov": [ { "bbox": { "b": 109.96807861328125, "coord_origin": "BOTTOMLEFT", "l": 107.4530029296875, "r": 504.384033203125, "t": 152.0926513671875 }, "charspan": [ 0, 283 ], "page_no": 18 } ], "self_ref": "#/texts/208", "text": "Sakti, S., E. Kelana, H. Riza, S. Sakai, K. Markov, and S. Nakamura (2008). Development of Indonesian large vocabulary continuous speech recognition system within a-STAR project. In Proceedings of the Workshop on Technologies and Corpora for Asia-Pacific Speech Translation (TCAST) ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Scheuerman, M. K., A. Hanna, and E. Denton (2021). Do datasets have politics? disciplinary values in computer vision dataset development. Proceedings of the ACM on Human-Computer Interaction 5 , 1-37.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 107.4646987915039, "r": 504.5961608886719, "t": 101.620849609375 }, "charspan": [ 0, 200 ], "page_no": 18 } ], "self_ref": "#/texts/209", "text": "Scheuerman, M. K., A. Hanna, and E. Denton (2021). Do datasets have politics? disciplinary values in computer vision dataset development. Proceedings of the ACM on Human-Computer Interaction 5 , 1-37." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "18", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 301.0190124511719, "r": 311.0133972167969, "t": 49.4544677734375 }, "charspan": [ 0, 2 ], "page_no": 18 } ], "self_ref": "#/texts/210", "text": "18" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "19", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.9600715637207, "coord_origin": "BOTTOMLEFT", "l": 301.01898193359375, "r": 311.10711669921875, "t": 49.34039306640625 }, "charspan": [ 0, 2 ], "page_no": 19 } ], "self_ref": "#/texts/211", "text": "19" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Appendix", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.158447265625, "coord_origin": "BOTTOMLEFT", "l": 107.39347076416016, "r": 157.5303955078125, "t": 718.9793090820312 }, "charspan": [ 0, 8 ], "page_no": 20 } ], "self_ref": "#/texts/212", "text": "Appendix" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "A Ethical Considerations and Broader Impacts Statement", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 681.8773193359375, "coord_origin": "BOTTOMLEFT", "l": 107.15185546875, "r": 411.63818359375, "t": 693.45751953125 }, "charspan": [ 0, 54 ], "page_no": 20 } ], "self_ref": "#/texts/213", "text": "A Ethical Considerations and Broader Impacts Statement" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As discussed in Section 1, the BigScience Research Workshop was conceived as a collaborative and value-driven endeavor from the start. All the ethical efforts were concentrated on implementing the values chosen first on the ethical charter and then on how to articulate those core values into specific ethical sensitive issues, such as data governance. This mechanism also allows ethical thinking to guide governance regarding technical matters. The articulation between the BigScience core values and those chosen by the collaborators contributing to data efforts was central. The importance of this collective exercise is due to the social impact that technologies such as LLMs have on the people impacted, directly and indirectly, positively and negatively. Moral exercises based on consensus, discussion around values, and how to link technical actions to ethical reflections is a strength that we believe is important within ML research. A critical analysis from an ethical perspective is fundamental to making different disciplines coexist in thinking around the social impact of these technologies and well define the object of analysis, as in this case, a multilingual dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 539.3563842773438, "coord_origin": "BOTTOMLEFT", "l": 107.05438995361328, "r": 505.24700927734375, "t": 669.4495849609375 }, "charspan": [ 0, 1185 ], "page_no": 20 } ], "self_ref": "#/texts/214", "text": "As discussed in Section 1, the BigScience Research Workshop was conceived as a collaborative and value-driven endeavor from the start. All the ethical efforts were concentrated on implementing the values chosen first on the ethical charter and then on how to articulate those core values into specific ethical sensitive issues, such as data governance. This mechanism also allows ethical thinking to guide governance regarding technical matters. The articulation between the BigScience core values and those chosen by the collaborators contributing to data efforts was central. The importance of this collective exercise is due to the social impact that technologies such as LLMs have on the people impacted, directly and indirectly, positively and negatively. Moral exercises based on consensus, discussion around values, and how to link technical actions to ethical reflections is a strength that we believe is important within ML research. A critical analysis from an ethical perspective is fundamental to making different disciplines coexist in thinking around the social impact of these technologies and well define the object of analysis, as in this case, a multilingual dataset." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "BigScience Values", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 515.703857421875, "coord_origin": "BOTTOMLEFT", "l": 107.67086791992188, "r": 184.2935791015625, "t": 525.5555419921875 }, "charspan": [ 0, 17 ], "page_no": 20 } ], "self_ref": "#/texts/215", "text": "BigScience Values" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Motivated by recent work on the values encoded in current approaches to research in NLP and ML more broadly (Leahy and Biderman, 2021; Birhane et al., 2021), which finds that narrow definitions of performance and efficiency were often prioritized over considerations of social impact in research and development. Even more relevant to the corpus creation aspect of our project, Scheuerman et al. (2021) outline how data efforts in computer vision tend to prioritize \" efficiency [over] care; universality [over] contextuality; impartiality [over] positionality . . . \". These ML research programs and systems in turn support the development of new technologies that carry these same values when deploying these technologies in production (Winner, 2017). This limits the potential positive societal benefits of the rapid advances of NLP research while increasing risks considerably.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 408.05853271484375, "coord_origin": "BOTTOMLEFT", "l": 107.15817260742188, "r": 505.6534118652344, "t": 505.49346923828125 }, "charspan": [ 0, 881 ], "page_no": 20 } ], "self_ref": "#/texts/216", "text": "Motivated by recent work on the values encoded in current approaches to research in NLP and ML more broadly (Leahy and Biderman, 2021; Birhane et al., 2021), which finds that narrow definitions of performance and efficiency were often prioritized over considerations of social impact in research and development. Even more relevant to the corpus creation aspect of our project, Scheuerman et al. (2021) outline how data efforts in computer vision tend to prioritize \" efficiency [over] care; universality [over] contextuality; impartiality [over] positionality . . . \". These ML research programs and systems in turn support the development of new technologies that carry these same values when deploying these technologies in production (Winner, 2017). This limits the potential positive societal benefits of the rapid advances of NLP research while increasing risks considerably." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Aware of these challenges, participants in BigScience collaboratively drafted an ethical charter 2 formalizing our core values and how they are articulated. It establishes the core values in order to allow its contributors to commit to them, both individually and collectively, and to ground discussions and choices made throughout the project in a common document. These values include notably openness and reproducibility as a scientific endeavor aimed at advancing the state of the art in a way that can be understood, interrogated, and re-used; responsibility of the participants to consider the social and legal context, and the social and environmental consequences of their work; and diversity and inclusivity . These last two are especially relevant to our data efforts, which aim to include text representative of diverse languages, varieties, and uses through a participatory approach to curation.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 304.11663818359375, "coord_origin": "BOTTOMLEFT", "l": 107.30754852294922, "r": 505.6373291015625, "t": 402.4932556152344 }, "charspan": [ 0, 907 ], "page_no": 20 } ], "self_ref": "#/texts/217", "text": "Aware of these challenges, participants in BigScience collaboratively drafted an ethical charter 2 formalizing our core values and how they are articulated. It establishes the core values in order to allow its contributors to commit to them, both individually and collectively, and to ground discussions and choices made throughout the project in a common document. These values include notably openness and reproducibility as a scientific endeavor aimed at advancing the state of the art in a way that can be understood, interrogated, and re-used; responsibility of the participants to consider the social and legal context, and the social and environmental consequences of their work; and diversity and inclusivity . These last two are especially relevant to our data efforts, which aim to include text representative of diverse languages, varieties, and uses through a participatory approach to curation." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Putting Our Values into Practice", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 280.4339294433594, "coord_origin": "BOTTOMLEFT", "l": 107.63008880615234, "r": 246.54986572265625, "t": 290.24237060546875 }, "charspan": [ 0, 32 ], "page_no": 20 } ], "self_ref": "#/texts/218", "text": "Putting Our Values into Practice" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Centering Participation in Data Curation Participatory approaches play a vital role in bridging the gaps between model development and deployment and in promoting fairness in ML applications (Rajkomar et al., 2018). They have received increased attention in recent years, with newer work calling to involve participants as full stake-holders of the entire research life-cycle rather to catering their role to post hoc model evaluation (Sloane et al., 2020; Caselli et al., 2021; Bondi et al., 2021), as exemplified by an organization like Maskhane (Nekoto et al., 2020) that brings together African researchers to collaboratively build NLP for African languages.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 195.07891845703125, "coord_origin": "BOTTOMLEFT", "l": 106.87773132324219, "r": 505.6535339355469, "t": 269.985107421875 }, "charspan": [ 0, 662 ], "page_no": 20 } ], "self_ref": "#/texts/219", "text": "Centering Participation in Data Curation Participatory approaches play a vital role in bridging the gaps between model development and deployment and in promoting fairness in ML applications (Rajkomar et al., 2018). They have received increased attention in recent years, with newer work calling to involve participants as full stake-holders of the entire research life-cycle rather to catering their role to post hoc model evaluation (Sloane et al., 2020; Caselli et al., 2021; Bondi et al., 2021), as exemplified by an organization like Maskhane (Nekoto et al., 2020) that brings together African researchers to collaboratively build NLP for African languages." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "With regard to developing LLMs, BigScience stands in contrast to previous work on models of similar size (Brown et al., 2020; Zhang et al., 2022) - where the majority of the development occurs in-house - by promoting engagement with other communities at every stage of the project from its design to the data curation to the eventual model training and release. Specifically, on the data curation aspect which is the focus of this paper, the involvement of a wide range of participants from various linguistic communities aims to help with the following aspects. First, Kreutzer et al. (2022) have shown in recent work that multilingual text data curation done without involving language-specific expertise leads to resources that are very different from the intentions of their creators, and these limitations carry on to the models trained on these datasets. Second, resources that are developed in collaboration with other communities are more likely to be more directly relevant to them, and thus to avoid reduce replication of model development by making the artifacts and tools we develop useful", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.51336669921875, "coord_origin": "BOTTOMLEFT", "l": 106.00700378417969, "r": 504.1680908203125, "t": 188.609619140625 }, "charspan": [ 0, 1101 ], "page_no": 20 } ], "self_ref": "#/texts/220", "text": "With regard to developing LLMs, BigScience stands in contrast to previous work on models of similar size (Brown et al., 2020; Zhang et al., 2022) - where the majority of the development occurs in-house - by promoting engagement with other communities at every stage of the project from its design to the data curation to the eventual model training and release. Specifically, on the data curation aspect which is the focus of this paper, the involvement of a wide range of participants from various linguistic communities aims to help with the following aspects. First, Kreutzer et al. (2022) have shown in recent work that multilingual text data curation done without involving language-specific expertise leads to resources that are very different from the intentions of their creators, and these limitations carry on to the models trained on these datasets. Second, resources that are developed in collaboration with other communities are more likely to be more directly relevant to them, and thus to avoid reduce replication of model development by making the artifacts and tools we develop useful" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "20", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.4928283691406, "r": 311.1434020996094, "t": 49.86578369140625 }, "charspan": [ 0, 2 ], "page_no": 20 } ], "self_ref": "#/texts/221", "text": "20" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "to more people and for more languages. Third, intentional curation and proper documentation of web-scale corpora takes a significant amount of human work and expertise, which can be distributed between a large number of participants in community efforts. Finally, community involvement can help foster trust and collective ownership of the artifacts we create.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 674.920654296875, "coord_origin": "BOTTOMLEFT", "l": 107.092529296875, "r": 504.2919006347656, "t": 717.5316772460938 }, "charspan": [ 0, 360 ], "page_no": 21 } ], "self_ref": "#/texts/222", "text": "to more people and for more languages. Third, intentional curation and proper documentation of web-scale corpora takes a significant amount of human work and expertise, which can be distributed between a large number of participants in community efforts. Finally, community involvement can help foster trust and collective ownership of the artifacts we create." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Addressing the Legal Landscape The legal status of webscraped datasets is extremely unclear in many jurisdictions, putting a substantial burden on both data creators and data users who wish to be involved with this process. While the principle of fair use generally protects academic researchers, it is not recognized in all jurisdictions and may not cover research carried out in an industry context. In consultation with our Legal Scholarship and Data Governance working groups, we developed a framework (Jernite et al., 2022) to uphold the rights and responsibilities of the many stakeholders in NLP data generation and collection, and provide assurances to downstream users as to how they are and are not authorized to use the dataset (Contractor et al., 2020).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 574.252685546875, "coord_origin": "BOTTOMLEFT", "l": 107.07137298583984, "r": 504.2640075683594, "t": 660.3369750976562 }, "charspan": [ 0, 765 ], "page_no": 21 } ], "self_ref": "#/texts/223", "text": "Addressing the Legal Landscape The legal status of webscraped datasets is extremely unclear in many jurisdictions, putting a substantial burden on both data creators and data users who wish to be involved with this process. While the principle of fair use generally protects academic researchers, it is not recognized in all jurisdictions and may not cover research carried out in an industry context. In consultation with our Legal Scholarship and Data Governance working groups, we developed a framework (Jernite et al., 2022) to uphold the rights and responsibilities of the many stakeholders in NLP data generation and collection, and provide assurances to downstream users as to how they are and are not authorized to use the dataset (Contractor et al., 2020)." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Limitations of the Approach.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 548.026611328125, "coord_origin": "BOTTOMLEFT", "l": 107.39588165283203, "r": 231.0724639892578, "t": 558.160888671875 }, "charspan": [ 0, 28 ], "page_no": 21 } ], "self_ref": "#/texts/224", "text": "Limitations of the Approach." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "While we believe that an approach grounded in community participation and prioritizing language expertise constitutes a promising step toward more responsible data curation and documentation, it still has important limitations. Among those, we primarily identify the use of data from the Common Crawl which represents a point of tension between our drive to present a research artifact that is comparable to previous work and values of consent and privacy (see Section 3). Our preprocessing removes some categories of PII but is still far from exhaustive, and the nature of crawled datasets makes it next to impossible to identify individual contributors and ask for their consent. Similar concerns apply to other existing NLP datasets we identified in the catalogue, including notably the WuDao web-based corpus (Yuan et al., 2021) which makes up a significant part of the Chinese language data. Additionally, while we hope that our intentional approach to selecting diverse data sources (mostly along axes of geographical diversity and domains) will lead to a more representative language dataset overall, our reliance on medium to large sources of digitized content still over-represents privileged voices and language varieties.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 396.7893371582031, "coord_origin": "BOTTOMLEFT", "l": 107.1647720336914, "r": 505.7449645996094, "t": 537.967041015625 }, "charspan": [ 0, 1232 ], "page_no": 21 } ], "self_ref": "#/texts/225", "text": "While we believe that an approach grounded in community participation and prioritizing language expertise constitutes a promising step toward more responsible data curation and documentation, it still has important limitations. Among those, we primarily identify the use of data from the Common Crawl which represents a point of tension between our drive to present a research artifact that is comparable to previous work and values of consent and privacy (see Section 3). Our preprocessing removes some categories of PII but is still far from exhaustive, and the nature of crawled datasets makes it next to impossible to identify individual contributors and ask for their consent. Similar concerns apply to other existing NLP datasets we identified in the catalogue, including notably the WuDao web-based corpus (Yuan et al., 2021) which makes up a significant part of the Chinese language data. Additionally, while we hope that our intentional approach to selecting diverse data sources (mostly along axes of geographical diversity and domains) will lead to a more representative language dataset overall, our reliance on medium to large sources of digitized content still over-represents privileged voices and language varieties." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "B Details on tools used to obtain crowdsourced dataset", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 367.3173522949219, "coord_origin": "BOTTOMLEFT", "l": 107.58988189697266, "r": 394.0401611328125, "t": 378.9547424316406 }, "charspan": [ 0, 54 ], "page_no": 21 } ], "self_ref": "#/texts/226", "text": "B Details on tools used to obtain crowdsourced dataset" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "B.1 Pseudocode to recreate the text structure from the HTML code", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 343.86578369140625, "coord_origin": "BOTTOMLEFT", "l": 107.50422668457031, "r": 399.44598388671875, "t": 353.5321350097656 }, "charspan": [ 0, 64 ], "page_no": 21 } ], "self_ref": "#/texts/227", "text": "B.1 Pseudocode to recreate the text structure from the HTML code" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The HTML code of a web page provides information about the structure of the text. The final structure of a web page is, however, the one produced by the rendering engine of the web browser and any CSS instructions. The latter two elements, which can vary enormously from one situation to another, always use the tag types for their rendering rules (Figure 9. Therefore, we have used a fairly simple heuristic on tag types to reconstruct the structure of the text extracted from an HTML code. To reconstruct the text, the HTML DOM, which can be represented as a tree (Figure 10), is traversed with an depth-first search algorithm. The text is initially empty and each time a new node with textual content is reached its content is concatenated according to the rules presented in the Algorithm 1. Block-type tags are for us: <address> , <article> , <aside> , <blockquote> , <body> , <br> , <button> , <canvas> , <caption> , <col> , <colgroup> , <dd> , <div> , <dl> , <dt> , <embed> , <fieldset> , <figcaption> , <figure> , <footer> , <form> , <h1> , <h2> , <h3> , <h4> , <h5> , <h6> , <header> , <hgroup> , <hr> , <li> , <map> , <noscript> , <object> , <ol> , <output> , <p> , <pre> , <progress> , <section> , <table> , <tbody> , <textarea> , <tfoot> , <th> , <thead> , <tr> , <ul> , and <video> . Inline-type tags are for us: <address> , <cite> , <details> , <datalist> , <iframe> , <img> , <input> , <label> , <legend> , <optgroup> , <q> , <select> , <summary> , <tbody> , <td> , and <time> .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 159.68008422851562, "coord_origin": "BOTTOMLEFT", "l": 105.97799682617188, "r": 505.5291442871094, "t": 333.0168151855469 }, "charspan": [ 0, 1493 ], "page_no": 21 } ], "self_ref": "#/texts/228", "text": "The HTML code of a web page provides information about the structure of the text. The final structure of a web page is, however, the one produced by the rendering engine of the web browser and any CSS instructions. The latter two elements, which can vary enormously from one situation to another, always use the tag types for their rendering rules (Figure 9. Therefore, we have used a fairly simple heuristic on tag types to reconstruct the structure of the text extracted from an HTML code. To reconstruct the text, the HTML DOM, which can be represented as a tree (Figure 10), is traversed with an depth-first search algorithm. The text is initially empty and each time a new node with textual content is reached its content is concatenated according to the rules presented in the Algorithm 1. Block-type tags are for us: <address> , <article> , <aside> , <blockquote> , <body> , <br> , <button> , <canvas> , <caption> , <col> , <colgroup> , <dd> , <div> , <dl> , <dt> , <embed> , <fieldset> , <figcaption> , <figure> , <footer> , <form> , <h1> , <h2> , <h3> , <h4> , <h5> , <h6> , <header> , <hgroup> , <hr> , <li> , <map> , <noscript> , <object> , <ol> , <output> , <p> , <pre> , <progress> , <section> , <table> , <tbody> , <textarea> , <tfoot> , <th> , <thead> , <tr> , <ul> , and <video> . Inline-type tags are for us: <address> , <cite> , <details> , <datalist> , <iframe> , <img> , <input> , <label> , <legend> , <optgroup> , <q> , <select> , <summary> , <tbody> , <td> , and <time> ." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "B.2 Visualization tool use cases", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 134.03282165527344, "coord_origin": "BOTTOMLEFT", "l": 107.45246124267578, "r": 246.79884338378906, "t": 143.6854248046875 }, "charspan": [ 0, 32 ], "page_no": 21 } ], "self_ref": "#/texts/229", "text": "B.2 Visualization tool use cases" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The visualisation tool was for us an iterative tool that we used to define new cleaning and filtering methods by visualising their effect on a subset of documents. This visualisation allowed us understand the impact of functions on the dataset at every stage of the processing pipeline (Figure 11 for advertisement detection for example), prompted us to adapt pipelines as well as introduce new functions for specific cases.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.44879150390625, "coord_origin": "BOTTOMLEFT", "l": 107.13905334472656, "r": 504.4093933105469, "t": 122.88519287109375 }, "charspan": [ 0, 424 ], "page_no": 21 } ], "self_ref": "#/texts/230", "text": "The visualisation tool was for us an iterative tool that we used to define new cleaning and filtering methods by visualising their effect on a subset of documents. This visualisation allowed us understand the impact of functions on the dataset at every stage of the processing pipeline (Figure 11 for advertisement detection for example), prompted us to adapt pipelines as well as introduce new functions for specific cases." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "21", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.40093994140625, "r": 310.9815673828125, "t": 49.91082763671875 }, "charspan": [ 0, 2 ], "page_no": 21 } ], "self_ref": "#/texts/231", "text": "21" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "(a) HTML code", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 658.0552368164062, "coord_origin": "BOTTOMLEFT", "l": 279.3606872558594, "r": 336.8915710449219, "t": 666.75927734375 }, "charspan": [ 0, 13 ], "page_no": 22 } ], "self_ref": "#/texts/232", "text": "(a) HTML code" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "(b) Web browser rendering", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 577.09423828125, "coord_origin": "BOTTOMLEFT", "l": 259.625732421875, "r": 356.66876220703125, "t": 586.33837890625 }, "charspan": [ 0, 25 ], "page_no": 22 } ], "self_ref": "#/texts/233", "text": "(b) Web browser rendering" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Figure 9: Example showing how a single line of HTML code is rendered by a browser's renderer. In this example, we can see that the tags <p> delimit different blocks which are therefore spaced by line breaks while other tags, such as <cite> , are rendered on the same line of text that precedes and follows them.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 526.5531005859375, "coord_origin": "BOTTOMLEFT", "l": 107.4919662475586, "r": 504.3517150878906, "t": 568.8838500976562 }, "charspan": [ 0, 311 ], "page_no": 22 } ], "self_ref": "#/texts/234", "text": "Figure 9: Example showing how a single line of HTML code is rendered by a browser's renderer. In this example, we can see that the tags <p> delimit different blocks which are therefore spaced by line breaks while other tags, such as <cite> , are rendered on the same line of text that precedes and follows them." }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 10: Simplified version of HTML DOM model on an example. Left: snippet of HTML code. Right: corresponding DOM. The yellow squares represent nodes with textual content.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 385.0390930175781, "coord_origin": "BOTTOMLEFT", "l": 107.6168441772461, "r": 505.73797607421875, "t": 405.4254455566406 }, "charspan": [ 0, 173 ], "page_no": 22 } ], "self_ref": "#/texts/235", "text": "Figure 10: Simplified version of HTML DOM model on an example. Left: snippet of HTML code. Right: corresponding DOM. The yellow squares represent nodes with textual content." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As a typical usage of the visualisation tool as a development tool, for documents coming from pseudo-crawls, we wanted to create a method to remove the parts of the documents that looked like a template, based on the principle that these templates would be identifiable by the fact that they would be repeated lines between documents. With the first version of the pipeline we could see from the estimates of the size of the final dataset (Figure 12) that a lot of content was removed. Looking at the examples (Figure 12c), we could confirm that a large part of the article text was removed. The cause of this behaviour was due to the fact that the same article was appearing at several different URLs as the website hierarchy had changed between the different common crawl dumps. For the final pipeline, we therefore added a custom deduplication of the urls as a first operation to target this change of addresses. With the final pipeline developed, less content was removed. By manually inspecting the examples, we could observe that the content removed from the documents was indeed the one initially targeted.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 229.16363525390625, "coord_origin": "BOTTOMLEFT", "l": 107.19416046142578, "r": 504.34423828125, "t": 358.9018249511719 }, "charspan": [ 0, 1113 ], "page_no": 22 } ], "self_ref": "#/texts/236", "text": "As a typical usage of the visualisation tool as a development tool, for documents coming from pseudo-crawls, we wanted to create a method to remove the parts of the documents that looked like a template, based on the principle that these templates would be identifiable by the fact that they would be repeated lines between documents. With the first version of the pipeline we could see from the estimates of the size of the final dataset (Figure 12) that a lot of content was removed. Looking at the examples (Figure 12c), we could confirm that a large part of the article text was removed. The cause of this behaviour was due to the fact that the same article was appearing at several different URLs as the website hierarchy had changed between the different common crawl dumps. For the final pipeline, we therefore added a custom deduplication of the urls as a first operation to target this change of addresses. With the final pipeline developed, less content was removed. By manually inspecting the examples, we could observe that the content removed from the documents was indeed the one initially targeted." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "B.3 Exhaustive list of functions used in (Crowd)Sourced dataset", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 201.75082397460938, "coord_origin": "BOTTOMLEFT", "l": 107.4989013671875, "r": 385.1993408203125, "t": 211.58331298828125 }, "charspan": [ 0, 63 ], "page_no": 22 } ], "self_ref": "#/texts/237", "text": "B.3 Exhaustive list of functions used in (Crowd)Sourced dataset" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We provide an exhaustive list of functions used in each of the processing pipeline for the crowdsourced dataset$^{14}$:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 169.5820770263672, "coord_origin": "BOTTOMLEFT", "l": 106.92504119873047, "r": 504.0096130371094, "t": 190.0667724609375 }, "charspan": [ 0, 119 ], "page_no": 22 } ], "self_ref": "#/texts/238", "text": "We provide an exhaustive list of functions used in each of the processing pipeline for the crowdsourced dataset$^{14}$:" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "replace_newline_with_space Takes in a batch of texts and for each text replaces the newline character \"", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 142.28408813476562, "coord_origin": "BOTTOMLEFT", "l": 107.48390197753906, "r": 505.6531982421875, "t": 163.11767578125 }, "charspan": [ 0, 103 ], "page_no": 22 } ], "self_ref": "#/texts/239", "text": "replace_newline_with_space Takes in a batch of texts and for each text replaces the newline character \"" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "n\" with a single space.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 131.00860595703125, "coord_origin": "BOTTOMLEFT", "l": 107.50057220458984, "r": 197.5637664794922, "t": 140.535888671875 }, "charspan": [ 0, 23 ], "page_no": 22 } ], "self_ref": "#/texts/240", "text": "n\" with a single space." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "remove_lines_with_code Takes in a batch of texts and removes lines with the following substrings: \"{\" , \"}\" , \"[if\" , \"<script\" ,", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 104.06524658203125, "coord_origin": "BOTTOMLEFT", "l": 107.94380187988281, "r": 505.38482666015625, "t": 124.65838623046875 }, "charspan": [ 0, 129 ], "page_no": 22 } ], "self_ref": "#/texts/241", "text": "remove_lines_with_code Takes in a batch of texts and removes lines with the following substrings: \"{\" , \"}\" , \"[if\" , \"<script\" ," }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "$^{14}$Code is available at preparation/blob/main/preprocessing/training/clean.py", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 70.0632553100586, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 302.90264892578125, "t": 88.0422134399414 }, "charspan": [ 0, 81 ], "page_no": 22 } ], "self_ref": "#/texts/242", "text": "$^{14}$Code is available at preparation/blob/main/preprocessing/training/clean.py" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "22", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.5227966308594, "r": 310.9815673828125, "t": 49.71807861328125 }, "charspan": [ 0, 2 ], "page_no": 22 } ], "self_ref": "#/texts/243", "text": "22" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "https://github.com/bigscience-workshop/data-", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 79.8563232421875, "coord_origin": "BOTTOMLEFT", "l": 337.4492492675781, "r": 505.48980712890625, "t": 88.86602783203125 }, "charspan": [ 0, 44 ], "page_no": 22 } ], "self_ref": "#/texts/244", "text": "https://github.com/bigscience-workshop/data-" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Algorithm 1 Pseudo-code to concatenate texts retrieved from the HTML DOM", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 695.1200561523438, "coord_origin": "BOTTOMLEFT", "l": 107.37068939208984, "r": 424.17236328125, "t": 704.909423828125 }, "charspan": [ 0, 72 ], "page_no": 23 } ], "self_ref": "#/texts/245", "text": "Algorithm 1 Pseudo-code to concatenate texts retrieved from the HTML DOM" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "1: text ← empty string", "parent": { "$ref": "#/groups/11" }, "prov": [ { "bbox": { "b": 679.9085083007812, "coord_origin": "BOTTOMLEFT", "l": 112.970458984375, "r": 207.52212524414062, "t": 689.6708984375 }, "charspan": [ 0, 22 ], "page_no": 23 } ], "self_ref": "#/texts/246", "text": "1: text ← empty string" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "2: 3: 4: 5: 6: 7: 8: 9: 10: 11: 12:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 560.2622680664062, "coord_origin": "BOTTOMLEFT", "l": 108.49598693847656, "r": 119.95504760742188, "t": 677.3692016601562 }, "charspan": [ 0, 35 ], "page_no": 23 } ], "self_ref": "#/texts/247", "text": "2: 3: 4: 5: 6: 7: 8: 9: 10: 11: 12:" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "for new _ text in list of texts retrieved by the DFS traversal do if new _ text is attached to a block-type tag then # Block elements are separated from the rest by a line break if text ends with a breaking line then text ← text + new _ text else if text ends with a space then text ← text without end space text ← text + breaking line + new _ text else text ← text + breaking line + new _ text end if", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 560.1168823242188, "coord_origin": "BOTTOMLEFT", "l": 124.93599700927734, "r": 465.58953857421875, "t": 678.2040405273438 }, "charspan": [ 0, 401 ], "page_no": 23 } ], "self_ref": "#/texts/248", "text": "for new _ text in list of texts retrieved by the DFS traversal do if new _ text is attached to a block-type tag then # Block elements are separated from the rest by a line break if text ends with a breaking line then text ← text + new _ text else if text ends with a space then text ← text without end space text ← text + breaking line + new _ text else text ← text + breaking line + new _ text end if" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "13: else if new _ text is attached to a inline-type tag then", "parent": { "$ref": "#/groups/12" }, "prov": [ { "bbox": { "b": 549.1381225585938, "coord_origin": "BOTTOMLEFT", "l": 108.49598693847656, "r": 350.57830810546875, "t": 558.7507934570312 }, "charspan": [ 0, 60 ], "page_no": 23 } ], "self_ref": "#/texts/249", "text": "13: else if new _ text is attached to a inline-type tag then" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "14:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 538.4442749023438, "coord_origin": "BOTTOMLEFT", "l": 108.4959716796875, "r": 119.95503234863281, "t": 546.4602661132812 }, "charspan": [ 0, 3 ], "page_no": 23 } ], "self_ref": "#/texts/250", "text": "14:" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "15: 16: 17: 18: 19: 20: 21: 22: 23:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 429.3533020019531, "coord_origin": "BOTTOMLEFT", "l": 108.4959716796875, "r": 119.95506286621094, "t": 524.6412353515625 }, "charspan": [ 0, 35 ], "page_no": 23 } ], "self_ref": "#/texts/251", "text": "15: 16: 17: 18: 19: 20: 21: 22: 23:" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "# space if text ends with a space or a breaking line then text ← text + new _ text else", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 494.6618347167969, "coord_origin": "BOTTOMLEFT", "l": 124.93599700927734, "r": 347.21533203125, "t": 547.0449829101562 }, "charspan": [ 0, 87 ], "page_no": 23 } ], "self_ref": "#/texts/252", "text": "# space if text ends with a space or a breaking line then text ← text + new _ text else" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Inline elements are separated from the rest by a line break or a", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 538.1683349609375, "coord_origin": "BOTTOMLEFT", "l": 169.33700561523438, "r": 503.998779296875, "t": 546.4671630859375 }, "charspan": [ 0, 64 ], "page_no": 23 } ], "self_ref": "#/texts/253", "text": "Inline elements are separated from the rest by a line break or a" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "text ← text + space + new _ text", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 483.68310546875, "coord_origin": "BOTTOMLEFT", "l": 169.54940795898438, "r": 303.6231689453125, "t": 493.3069763183594 }, "charspan": [ 0, 32 ], "page_no": 23 } ], "self_ref": "#/texts/254", "text": "text ← text + space + new _ text" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "end if else text ← text + new _ text end if end for", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 429.2078552246094, "coord_origin": "BOTTOMLEFT", "l": 124.93596649169922, "r": 255.95217895507812, "t": 481.80023193359375 }, "charspan": [ 0, 51 ], "page_no": 23 } ], "self_ref": "#/texts/255", "text": "end if else text ← text + new _ text end if end for" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Old text", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 382.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 110.66666412353516, "r": 142.6666717529297, "t": 392.3333435058594 }, "charspan": [ 0, 8 ], "page_no": 23 } ], "self_ref": "#/texts/256", "text": "Old text" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "New text", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 382.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 307.3333435058594, "r": 344, "t": 393 }, "charspan": [ 0, 8 ], "page_no": 23 } ], "self_ref": "#/texts/257", "text": "New text" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 11: Example of showing sample changes throughout each step of the processing pipeline. In the following example, users can notice that advertisement text were removed from the main article.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 85.71142578125, "coord_origin": "BOTTOMLEFT", "l": 107.49712371826172, "r": 504.13763427734375, "t": 106.47540283203125 }, "charspan": [ 0, 196 ], "page_no": 23 } ], "self_ref": "#/texts/258", "text": "Figure 11: Example of showing sample changes throughout each step of the processing pipeline. In the following example, users can notice that advertisement text were removed from the main article." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "23", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.39337158203125, "r": 310.9815979003906, "t": 49.5499267578125 }, "charspan": [ 0, 2 ], "page_no": 23 } ], "self_ref": "#/texts/259", "text": "23" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The purpose of this application is to sequentially view the changes made to dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707, "coord_origin": "BOTTOMLEFT", "l": 111.99247741699219, "r": 298.3333435058594, "t": 715.3333129882812 }, "charspan": [ 0, 84 ], "page_no": 24 } ], "self_ref": "#/texts/260", "text": "The purpose of this application is to sequentially view the changes made to dataset." }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "(a) Pipeline v0", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 590.6058959960938, "coord_origin": "BOTTOMLEFT", "l": 281.3626708984375, "r": 335.3610534667969, "t": 599.201171875 }, "charspan": [ 0, 15 ], "page_no": 24 } ], "self_ref": "#/texts/261", "text": "(a) Pipeline v0" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The purpose of this application is to sequentially view the changes made to dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 571, "coord_origin": "BOTTOMLEFT", "l": 111.15206909179688, "r": 297.6666564941406, "t": 577.7601318359375 }, "charspan": [ 0, 84 ], "page_no": 24 } ], "self_ref": "#/texts/262", "text": "The purpose of this application is to sequentially view the changes made to dataset." }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "(b) Pipeline v2", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 453.2032775878906, "coord_origin": "BOTTOMLEFT", "l": 281.20111083984375, "r": 335.15643310546875, "t": 462.137451171875 }, "charspan": [ 0, 15 ], "page_no": 24 } ], "self_ref": "#/texts/263", "text": "(b) Pipeline v2" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "(c) Sample example difference between pipeline versions", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 153.7512664794922, "coord_origin": "BOTTOMLEFT", "l": 203.77529907226562, "r": 409.4661865234375, "t": 162.4075927734375 }, "charspan": [ 0, 55 ], "page_no": 24 } ], "self_ref": "#/texts/264", "text": "(c) Sample example difference between pipeline versions" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Figure 12: High level statistics between two seperate pipelines and a sample example of the difference between two pipelines. First iteration (Figure 12a) generated a 7Mb dataset. After some careful tweaking, and some observed samples, we proposed a new pipeline in order to preserve previously wrongly removed data (Figure 12b) which generated a 134Mb dataset (x18). A example sample is available in Figure 12c", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 92.30107879638672, "coord_origin": "BOTTOMLEFT", "l": 107.11819458007812, "r": 504.3509521484375, "t": 145.46380615234375 }, "charspan": [ 0, 411 ], "page_no": 24 } ], "self_ref": "#/texts/265", "text": "Figure 12: High level statistics between two seperate pipelines and a sample example of the difference between two pipelines. First iteration (Figure 12a) generated a 7Mb dataset. After some careful tweaking, and some observed samples, we proposed a new pipeline in order to preserve previously wrongly removed data (Figure 12b) which generated a 134Mb dataset (x18). A example sample is available in Figure 12c" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "24", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.9600830078125, "coord_origin": "BOTTOMLEFT", "l": 300.4591064453125, "r": 310.9815673828125, "t": 49.365234375 }, "charspan": [ 0, 2 ], "page_no": 24 } ], "self_ref": "#/texts/266", "text": "24" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "remove_html_spans Takes in a batch of texts and removes lines with the following substrings: \"<span\" , \"</span>\" , \"<div\" , \"<a\" , \"</div>\" , \"</a>\" , \"br>\" ,", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.8570556640625, "coord_origin": "BOTTOMLEFT", "l": 107.89938354492188, "r": 505.3880920410156, "t": 717.5349731445312 }, "charspan": [ 0, 158 ], "page_no": 25 } ], "self_ref": "#/texts/267", "text": "remove_html_spans Takes in a batch of texts and removes lines with the following substrings: \"<span\" , \"</span>\" , \"<div\" , \"<a\" , \"</div>\" , \"</a>\" , \"br>\" ," }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "remove_html_spans_sanad Takes in a batch of texts and removes lines with the following substrings: \"<img\" , \"]]>\" , \"<![CDATA\" , \"//DW\" , \"var \" , \"xtImg\" , \"To view this video please enable JavaScript\" ,", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 669.6586303710938, "coord_origin": "BOTTOMLEFT", "l": 107.93336486816406, "r": 505.38201904296875, "t": 690.0968627929688 }, "charspan": [ 0, 204 ], "page_no": 25 } ], "self_ref": "#/texts/268", "text": "remove_html_spans_sanad Takes in a batch of texts and removes lines with the following substrings: \"<img\" , \"]]>\" , \"<![CDATA\" , \"//DW\" , \"var \" , \"xtImg\" , \"To view this video please enable JavaScript\" ," }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "remove_wiki_mojibake Takes in a batch of texts and removes lines with the following substrings: \"À À\"", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 641.5042114257812, "coord_origin": "BOTTOMLEFT", "l": 107.98168182373047, "r": 505.3809814453125, "t": 662.6681518554688 }, "charspan": [ 0, 101 ], "page_no": 25 } ], "self_ref": "#/texts/269", "text": "remove_wiki_mojibake Takes in a batch of texts and removes lines with the following substrings: \"À À\"" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "strip_substrings_en_wiktionary Takes in a batch of texts and removes the following substrings:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 624.7998046875, "coord_origin": "BOTTOMLEFT", "l": 107.56604766845703, "r": 495.7628173828125, "t": 634.7673950195312 }, "charspan": [ 0, 94 ], "page_no": 25 } ], "self_ref": "#/texts/270", "text": "strip_substrings_en_wiktionary Takes in a batch of texts and removes the following substrings:" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· This entry needs pronunciation information", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 604.3540649414062, "coord_origin": "BOTTOMLEFT", "l": 134.8498992919922, "r": 317.2142333984375, "t": 614.2539672851562 }, "charspan": [ 0, 44 ], "page_no": 25 } ], "self_ref": "#/texts/271", "text": "· This entry needs pronunciation information" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Please try to find a suitable image on Wikimedia Commons or upload one there yourself!This entry need pronunciation information", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 578.218017578125, "coord_origin": "BOTTOMLEFT", "l": 135.08482360839844, "r": 503.9967956542969, "t": 598.8729858398438 }, "charspan": [ 0, 129 ], "page_no": 25 } ], "self_ref": "#/texts/272", "text": "· Please try to find a suitable image on Wikimedia Commons or upload one there yourself!This entry need pronunciation information" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· You may continue to edit this entry while the discussion proceeds, but please mention significant edits at the RFD discussion and ensure that the intention of votes already cast is not left unclear", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 541.2249145507812, "coord_origin": "BOTTOMLEFT", "l": 134.6595916748047, "r": 504.0032653808594, "t": 572.867919921875 }, "charspan": [ 0, 199 ], "page_no": 25 } ], "self_ref": "#/texts/273", "text": "· You may continue to edit this entry while the discussion proceeds, but please mention significant edits at the RFD discussion and ensure that the intention of votes already cast is not left unclear" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· This entry is part of the phrasebook project, which presents criteria for inclusion based on utility, simplicity and commonality", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 515.1671142578125, "coord_origin": "BOTTOMLEFT", "l": 134.80784606933594, "r": 504.00396728515625, "t": 535.423828125 }, "charspan": [ 0, 130 ], "page_no": 25 } ], "self_ref": "#/texts/274", "text": "· This entry is part of the phrasebook project, which presents criteria for inclusion based on utility, simplicity and commonality" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· If you are a native speaker with a microphone, please record some and upload them", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 499.58172607421875, "coord_origin": "BOTTOMLEFT", "l": 134.9659423828125, "r": 477.75262451171875, "t": 509.5697326660156 }, "charspan": [ 0, 83 ], "page_no": 25 } ], "self_ref": "#/texts/275", "text": "· If you are a native speaker with a microphone, please record some and upload them" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· If you are familiar with the IPA then please add some!", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 484.293701171875, "coord_origin": "BOTTOMLEFT", "l": 134.95401000976562, "r": 360.0334777832031, "t": 494.2102966308594 }, "charspan": [ 0, 56 ], "page_no": 25 } ], "self_ref": "#/texts/276", "text": "· If you are familiar with the IPA then please add some!" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Feel free to edit this entry as normal, but do not remove rfv until the request has been resolved", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 458.76220703125, "coord_origin": "BOTTOMLEFT", "l": 134.93235778808594, "r": 504.0021667480469, "t": 478.83172607421875 }, "charspan": [ 0, 99 ], "page_no": 25 } ], "self_ref": "#/texts/277", "text": "· Feel free to edit this entry as normal, but do not remove rfv until the request has been resolved" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· This entry needs quotations to illustrate usage", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 442.86590576171875, "coord_origin": "BOTTOMLEFT", "l": 134.72317504882812, "r": 328.0805969238281, "t": 452.72393798828125 }, "charspan": [ 0, 49 ], "page_no": 25 } ], "self_ref": "#/texts/278", "text": "· This entry needs quotations to illustrate usage" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· If you are familiar with the IPA then please add some!This entry needs audio files", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 427.495849609375, "coord_origin": "BOTTOMLEFT", "l": 134.92396545410156, "r": 468.6138916015625, "t": 437.33489990234375 }, "charspan": [ 0, 84 ], "page_no": 25 } ], "self_ref": "#/texts/279", "text": "· If you are familiar with the IPA then please add some!This entry needs audio files" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Please see that page for discussion and justifications", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 412.8950500488281, "coord_origin": "BOTTOMLEFT", "l": 134.7511444091797, "r": 354.2661437988281, "t": 422.4875793457031 }, "charspan": [ 0, 56 ], "page_no": 25 } ], "self_ref": "#/texts/280", "text": "· Please see that page for discussion and justifications" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· If you are familiar with the IPA or enPR then please add some!A user has added this entry to requests for verification(+) If it cannot be verified that this term meets our attestation criteria, it will be deleted", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 376.0942077636719, "coord_origin": "BOTTOMLEFT", "l": 134.6541290283203, "r": 504.0920104980469, "t": 407.43853759765625 }, "charspan": [ 0, 214 ], "page_no": 25 } ], "self_ref": "#/texts/281", "text": "· If you are familiar with the IPA or enPR then please add some!A user has added this entry to requests for verification(+) If it cannot be verified that this term meets our attestation criteria, it will be deleted" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· This entry needs a photograph or drawing for illustration", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 360.15167236328125, "coord_origin": "BOTTOMLEFT", "l": 134.82998657226562, "r": 373.03564453125, "t": 370.11505126953125 }, "charspan": [ 0, 59 ], "page_no": 25 } ], "self_ref": "#/texts/282", "text": "· This entry needs a photograph or drawing for illustration" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· A user has added this entry to requests for deletion(+)", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 344.77435302734375, "coord_origin": "BOTTOMLEFT", "l": 134.73095703125, "r": 360.253662109375, "t": 354.7396240234375 }, "charspan": [ 0, 57 ], "page_no": 25 } ], "self_ref": "#/texts/283", "text": "· A user has added this entry to requests for deletion(+)" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Do not remove the rfd until the debate has finished", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 330.2142028808594, "coord_origin": "BOTTOMLEFT", "l": 134.82054138183594, "r": 346.0554504394531, "t": 339.944580078125 }, "charspan": [ 0, 53 ], "page_no": 25 } ], "self_ref": "#/texts/284", "text": "· Do not remove the rfd until the debate has finished" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· This entry needs audio files", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 314.7401428222656, "coord_origin": "BOTTOMLEFT", "l": 134.74618530273438, "r": 252.8966827392578, "t": 324.67437744140625 }, "charspan": [ 0, 30 ], "page_no": 25 } ], "self_ref": "#/texts/285", "text": "· This entry needs audio files" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· If you come across any interesting, durably archived quotes then please add them!This entry is part of the phrasebook project, which presents criteria for inclusion based on utility, simplicity and commonality", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 277.9161376953125, "coord_origin": "BOTTOMLEFT", "l": 134.7951202392578, "r": 505.7423400878906, "t": 309.52691650390625 }, "charspan": [ 0, 211 ], "page_no": 25 } ], "self_ref": "#/texts/286", "text": "· If you come across any interesting, durably archived quotes then please add them!This entry is part of the phrasebook project, which presents criteria for inclusion based on utility, simplicity and commonality" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· (For audio required quickly, visit WT:APR)", "parent": { "$ref": "#/groups/13" }, "prov": [ { "bbox": { "b": 262.127685546875, "coord_origin": "BOTTOMLEFT", "l": 134.90476989746094, "r": 315.1131286621094, "t": 272.11083984375 }, "charspan": [ 0, 44 ], "page_no": 25 } ], "self_ref": "#/texts/287", "text": "· (For audio required quickly, visit WT:APR)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "remove_references_{lang} Removes lines that do not contain a minimum ratio of stopwords, as defined for each language$^{15}$. Note, currently does not support languages with different segmentation (e.g. Chinese). Designed for academic datasets.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 220.374755859375, "coord_origin": "BOTTOMLEFT", "l": 107.2276611328125, "r": 504.1107177734375, "t": 252.36590576171875 }, "charspan": [ 0, 244 ], "page_no": 25 } ], "self_ref": "#/texts/288", "text": "remove_references_{lang} Removes lines that do not contain a minimum ratio of stopwords, as defined for each language$^{15}$. Note, currently does not support languages with different segmentation (e.g. Chinese). Designed for academic datasets." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "split_sentences_{lang} Builds a sentence splitter depending on the language passed: For Arabic, Catalan, Basque, Indonesian, and Chinese (both simplified and traditional), we use the Stanza tokenizer (Qi et al., 2020). For English, French, Portuguese, and Spanish, we use the NLTK tokenizer (Bird et al., 2009). For Bengalic, Gujarati, Hindi, Kannada, Malayalam, Marathi, Punjabi, Tamil, and Telugu, we use the Indic NLP library tokenizer (Kunchukuttan, 2020). For Vietnamese, we use the Underthesea tokenizer $^{16}$.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 149.486083984375, "coord_origin": "BOTTOMLEFT", "l": 107.14425659179688, "r": 505.24224853515625, "t": 213.59161376953125 }, "charspan": [ 0, 518 ], "page_no": 25 } ], "self_ref": "#/texts/289", "text": "split_sentences_{lang} Builds a sentence splitter depending on the language passed: For Arabic, Catalan, Basque, Indonesian, and Chinese (both simplified and traditional), we use the Stanza tokenizer (Qi et al., 2020). For English, French, Portuguese, and Spanish, we use the NLTK tokenizer (Bird et al., 2009). For Bengalic, Gujarati, Hindi, Kannada, Malayalam, Marathi, Punjabi, Tamil, and Telugu, we use the Indic NLP library tokenizer (Kunchukuttan, 2020). For Vietnamese, we use the Underthesea tokenizer $^{16}$." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "filter_remove_empty_docs Removes documents that have a length of 0 when whitespace is removed.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 132.865478515625, "coord_origin": "BOTTOMLEFT", "l": 107.44434356689453, "r": 505.7386779785156, "t": 142.80853271484375 }, "charspan": [ 0, 94 ], "page_no": 25 } ], "self_ref": "#/texts/290", "text": "filter_remove_empty_docs Removes documents that have a length of 0 when whitespace is removed." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "filter_wiki_user_titles Removes documents where the Wikimedia metadata title starts with \"user\" ,", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 116.52423095703125, "coord_origin": "BOTTOMLEFT", "l": 107.54591369628906, "r": 505.1902160644531, "t": 126.29852294921875 }, "charspan": [ 0, 97 ], "page_no": 25 } ], "self_ref": "#/texts/291", "text": "filter_wiki_user_titles Removes documents where the Wikimedia metadata title starts with \"user\" ," }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "filter_wiki_non_text_type Removes documents where the Wikimedia metadata type is not \"text\"", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 100.00848388671875, "coord_origin": "BOTTOMLEFT", "l": 107.7824935913086, "r": 498.41412353515625, "t": 110.23992919921875 }, "charspan": [ 0, 91 ], "page_no": 25 } ], "self_ref": "#/texts/292", "text": "filter_wiki_non_text_type Removes documents where the Wikimedia metadata type is not \"text\"" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{15}$https://github.com/bigscience-workshop/catalogue_data/blob/master/clean_helpers/stopwords.py", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 79.85125732421875, "coord_origin": "BOTTOMLEFT", "l": 118.72645568847656, "r": 472.96856689453125, "t": 90.87200927734375 }, "charspan": [ 0, 99 ], "page_no": 25 } ], "self_ref": "#/texts/293", "text": "$^{15}$https://github.com/bigscience-workshop/catalogue_data/blob/master/clean_helpers/stopwords.py" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{16}$https://github.com/undertheseanlp/underthesea", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 70.01397705078125, "coord_origin": "BOTTOMLEFT", "l": 119.03575897216797, "r": 293.2182312011719, "t": 79.35113525390625 }, "charspan": [ 0, 52 ], "page_no": 25 } ], "self_ref": "#/texts/294", "text": "$^{16}$https://github.com/undertheseanlp/underthesea" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "25", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960086822509766, "coord_origin": "BOTTOMLEFT", "l": 300.4989929199219, "r": 310.9815673828125, "t": 49.72979736328125 }, "charspan": [ 0, 2 ], "page_no": 25 } ], "self_ref": "#/texts/295", "text": "25" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "filter_small_docs Discards documents with less than 15 words. Tokenization is done via whitespace tokenizer.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.97607421875, "coord_origin": "BOTTOMLEFT", "l": 107.3839340209961, "r": 504, "t": 717.418701171875 }, "charspan": [ 0, 108 ], "page_no": 26 } ], "self_ref": "#/texts/296", "text": "filter_small_docs Discards documents with less than 15 words. Tokenization is done via whitespace tokenizer." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "filter_small_docs_bytes_{i} Discards documents with less than either 300 or 1024 bytes of text", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 680.2073364257812, "coord_origin": "BOTTOMLEFT", "l": 107.50458526611328, "r": 490.8549499511719, "t": 690.1724853515625 }, "charspan": [ 0, 94 ], "page_no": 26 } ], "self_ref": "#/texts/297", "text": "filter_small_docs_bytes_{i} Discards documents with less than either 300 or 1024 bytes of text" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dedup_template_soft Removes lines that are a minimum of 15 characters long and occur 10 or more times.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 653.2901000976562, "coord_origin": "BOTTOMLEFT", "l": 107.46633911132812, "r": 504.0035705566406, "t": 674.1968383789062 }, "charspan": [ 0, 102 ], "page_no": 26 } ], "self_ref": "#/texts/298", "text": "dedup_template_soft Removes lines that are a minimum of 15 characters long and occur 10 or more times." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dedup_pseudocrawl_newspapers Removes lines that occur 2 or more times.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 636.865478515625, "coord_origin": "BOTTOMLEFT", "l": 107.46752166748047, "r": 418.4736022949219, "t": 646.830810546875 }, "charspan": [ 0, 70 ], "page_no": 26 } ], "self_ref": "#/texts/299", "text": "dedup_pseudocrawl_newspapers Removes lines that occur 2 or more times." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dedup_document Removes duplicate documents ignoring whitespaces and punctuation so only keeping characters and keeps one occurrence.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 609.6040649414062, "coord_origin": "BOTTOMLEFT", "l": 107.54133605957031, "r": 504.35089111328125, "t": 630.2672729492188 }, "charspan": [ 0, 132 ], "page_no": 26 } ], "self_ref": "#/texts/300", "text": "dedup_document Removes duplicate documents ignoring whitespaces and punctuation so only keeping characters and keeps one occurrence." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dedup_document_on_url Removes duplicate documents based on matched url while ignoring query parameters and keeps one occurrence.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 582.0811767578125, "coord_origin": "BOTTOMLEFT", "l": 107.38944244384766, "r": 504.3499755859375, "t": 602.795654296875 }, "charspan": [ 0, 128 ], "page_no": 26 } ], "self_ref": "#/texts/301", "text": "dedup_document_on_url Removes duplicate documents based on matched url while ignoring query parameters and keeps one occurrence." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dedup_document_on_url_lm_es_pseudocrawl-filtered_341_es_cointelegraph_com Removes duplicate documents based on the normalized urls (e.g., $URL and $URL/amp are treated as the same) without the query parameters and keeps one occurrence.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 543.7072143554688, "coord_origin": "BOTTOMLEFT", "l": 107.33811950683594, "r": 504.1123352050781, "t": 575.7845458984375 }, "charspan": [ 0, 235 ], "page_no": 26 } ], "self_ref": "#/texts/302", "text": "dedup_document_on_url_lm_es_pseudocrawl-filtered_341_es_cointelegraph_com Removes duplicate documents based on the normalized urls (e.g., $URL and $URL/amp are treated as the same) without the query parameters and keeps one occurrence." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dedup_document_on_url_lm_en_pseudocrawl_filtered_619_www_qut_edu_au Removes duplicate documents based on the url without query parameters except for the \"id\" and \"new-id\" query parameters. The \"new-id\" query parameter is changed into a simple \"id\" parameter.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 505.33740234375, "coord_origin": "BOTTOMLEFT", "l": 107.29246520996094, "r": 505.6521301269531, "t": 537.4888916015625 }, "charspan": [ 0, 258 ], "page_no": 26 } ], "self_ref": "#/texts/303", "text": "dedup_document_on_url_lm_en_pseudocrawl_filtered_619_www_qut_edu_au Removes duplicate documents based on the url without query parameters except for the \"id\" and \"new-id\" query parameters. The \"new-id\" query parameter is changed into a simple \"id\" parameter." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "concatenate_lm_fr_ester Concatenate the text sorted by the id number in the metadata.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 489.0220642089844, "coord_origin": "BOTTOMLEFT", "l": 107.48674774169922, "r": 459.2314453125, "t": 498.90399169921875 }, "charspan": [ 0, 85 ], "page_no": 26 } ], "self_ref": "#/texts/304", "text": "concatenate_lm_fr_ester Concatenate the text sorted by the id number in the metadata." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "C Exhaustive list of human curated filters used on OSCAR", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 460.21539306640625, "coord_origin": "BOTTOMLEFT", "l": 107.63353729248047, "r": 415.9540100097656, "t": 471.8385925292969 }, "charspan": [ 0, 56 ], "page_no": 26 } ], "self_ref": "#/texts/305", "text": "C Exhaustive list of human curated filters used on OSCAR" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Before performing the filtering step, we did a cleaning step to modify the documents by standardizing whitespace and removing links, non-printable characters, and long words beyond a character threshold. These steps were designed to remove \"non natural\" language parts of the document (i.e. texts that are machine generated or not language, such as URLs).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 403.96820068359375, "coord_origin": "BOTTOMLEFT", "l": 106.82133483886719, "r": 505.741455078125, "t": 446.9094543457031 }, "charspan": [ 0, 355 ], "page_no": 26 } ], "self_ref": "#/texts/306", "text": "Before performing the filtering step, we did a cleaning step to modify the documents by standardizing whitespace and removing links, non-printable characters, and long words beyond a character threshold. These steps were designed to remove \"non natural\" language parts of the document (i.e. texts that are machine generated or not language, such as URLs)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Many of these filters require to split a document into words. For Chinese, we used the SentencePiece unigram tokenizer. For Vietnamese, since a word can be composed of two or three sub-words separated by spaces, we augmented the list of space separated tokens by the list of two and three consecutive space separated tokens.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 354.909423828125, "coord_origin": "BOTTOMLEFT", "l": 107.27851104736328, "r": 504.0705261230469, "t": 397.4911804199219 }, "charspan": [ 0, 324 ], "page_no": 26 } ], "self_ref": "#/texts/307", "text": "Many of these filters require to split a document into words. For Chinese, we used the SentencePiece unigram tokenizer. For Vietnamese, since a word can be composed of two or three sub-words separated by spaces, we augmented the list of space separated tokens by the list of two and three consecutive space separated tokens." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on number of words We discarded documents with too few words, as they often contain incorrect sentences, or contain no context for a model to learn correctly.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 320.3271484375, "coord_origin": "BOTTOMLEFT", "l": 107.52661895751953, "r": 504.0008850097656, "t": 341.52886962890625 }, "charspan": [ 0, 165 ], "page_no": 26 } ], "self_ref": "#/texts/308", "text": "Filter on number of words We discarded documents with too few words, as they often contain incorrect sentences, or contain no context for a model to learn correctly." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on character repetition ratio To remove documents containing many repetitions, for a given n (determined in practice according to the language by native speakers), we counted the occurrence of each character n -gram present in the document. We defined the character repetition ratio as the ratio of the sum of the k largest occurrences by the sum of all occurrences, and we discarded documents with a too high ratio.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 253.3970947265625, "coord_origin": "BOTTOMLEFT", "l": 107.12351989746094, "r": 504.0930480957031, "t": 306.98614501953125 }, "charspan": [ 0, 423 ], "page_no": 26 } ], "self_ref": "#/texts/309", "text": "Filter on character repetition ratio To remove documents containing many repetitions, for a given n (determined in practice according to the language by native speakers), we counted the occurrence of each character n -gram present in the document. We defined the character repetition ratio as the ratio of the sum of the k largest occurrences by the sum of all occurrences, and we discarded documents with a too high ratio." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "If k = 1 , short sentences are much more likely to have a high character repetition ratio, since the most frequent n -gram represents a larger proportion of the sentence. If k is the number of occurrences greater than or equal to 2 , very long documents, but not necessarily including repetitions, tend to have a high character repetition ratio, since these texts inherently have a wide diversity of n -grams. We found that k = ⌊ √ N ⌋ , with N the number of different n -grams found in the document, counterbalances well this effect in practice.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 186.1220703125, "coord_origin": "BOTTOMLEFT", "l": 107.15222930908203, "r": 505.2486572265625, "t": 251.90753173828125 }, "charspan": [ 0, 546 ], "page_no": 26 } ], "self_ref": "#/texts/310", "text": "If k = 1 , short sentences are much more likely to have a high character repetition ratio, since the most frequent n -gram represents a larger proportion of the sentence. If k is the number of occurrences greater than or equal to 2 , very long documents, but not necessarily including repetitions, tend to have a high character repetition ratio, since these texts inherently have a wide diversity of n -grams. We found that k = ⌊ √ N ⌋ , with N the number of different n -grams found in the document, counterbalances well this effect in practice." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Example: Take the sentence \" ok_ok_good_ok \" and n = 3 . Character n -grams, with their frequencies, are given in the following table.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 158.8583984375, "coord_origin": "BOTTOMLEFT", "l": 107.02708435058594, "r": 505.2845153808594, "t": 179.47991943359375 }, "charspan": [ 0, 134 ], "page_no": 26 } ], "self_ref": "#/texts/311", "text": "Example: Take the sentence \" ok_ok_good_ok \" and n = 3 . Character n -grams, with their frequencies, are given in the following table." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Since we have 9 different character n -grams, N = 9 and k = ⌊ √ N ⌋ = 3 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 96.54864501953125, "coord_origin": "BOTTOMLEFT", "l": 107.19706726074219, "r": 401.9629821777344, "t": 115.20494079589844 }, "charspan": [ 0, 73 ], "page_no": 26 } ], "self_ref": "#/texts/312", "text": "Since we have 9 different character n -grams, N = 9 and k = ⌊ √ N ⌋ = 3 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The sum of the k largest occurrences is 2 + 2 + 1 = 5 and the sum of all occurrences is 11 . Thus, the character repetition ratio for this sentence is 5 $_{11}$.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.4072265625, "coord_origin": "BOTTOMLEFT", "l": 107.0120620727539, "r": 503.99993896484375, "t": 90.19830322265625 }, "charspan": [ 0, 161 ], "page_no": 26 } ], "self_ref": "#/texts/313", "text": "The sum of the k largest occurrences is 2 + 2 + 1 = 5 and the sum of all occurrences is 11 . Thus, the character repetition ratio for this sentence is 5 $_{11}$." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "26", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.4974365234375, "r": 311.2148742675781, "t": 49.81683349609375 }, "charspan": [ 0, 2 ], "page_no": 26 } ], "self_ref": "#/texts/314", "text": "26" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on word repetition ratio As a complement to the previous filter, we remove documents that have commonly repeated similar long sentences. More specifically, we create a filter for the repetitions by looking this time at the occurrences of the word n -grams, for a chosen n parameter. We define the word repetition ratio as the ratio of the sum of the occurrences greater than or equal to 2 to the sum of all occurrences, and we discard documents with too high of a ratio. Contrary to the filter on the character repetition ratios, we did not find a bias of this method giving systematically higher or lower scores to longer or short documents. This filter is more robust in finding documents with long exact duplicated sentences in them, while the previous one is used to find short to medium sized repetitions.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 620.4337158203125, "coord_origin": "BOTTOMLEFT", "l": 107.1661605834961, "r": 505.7427978515625, "t": 717.64697265625 }, "charspan": [ 0, 817 ], "page_no": 27 } ], "self_ref": "#/texts/315", "text": "Filter on word repetition ratio As a complement to the previous filter, we remove documents that have commonly repeated similar long sentences. More specifically, we create a filter for the repetitions by looking this time at the occurrences of the word n -grams, for a chosen n parameter. We define the word repetition ratio as the ratio of the sum of the occurrences greater than or equal to 2 to the sum of all occurrences, and we discard documents with too high of a ratio. Contrary to the filter on the character repetition ratios, we did not find a bias of this method giving systematically higher or lower scores to longer or short documents. This filter is more robust in finding documents with long exact duplicated sentences in them, while the previous one is used to find short to medium sized repetitions." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on special character ratio We established a list of special characters, including emojis, and simply discard documents with a special character ratio above a certain threshold.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 586.250244140625, "coord_origin": "BOTTOMLEFT", "l": 107.4928207397461, "r": 504.02972412109375, "t": 606.7445678710938 }, "charspan": [ 0, 183 ], "page_no": 27 } ], "self_ref": "#/texts/316", "text": "Filter on special character ratio We established a list of special characters, including emojis, and simply discard documents with a special character ratio above a certain threshold." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on closed class word ratio We found that having a low closed class word ratio in a document was one of the best indicators of a non-human generated content. We built lists of closed class words for each language by taking pre-existing lists, for example from Universal Dependencies$^{17}$, which were then reviewed by native speakers. We discard documents with a too low closed class word ratio.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 530.1297607421875, "coord_origin": "BOTTOMLEFT", "l": 107.13435363769531, "r": 505.7391662597656, "t": 572.8287353515625 }, "charspan": [ 0, 402 ], "page_no": 27 } ], "self_ref": "#/texts/317", "text": "Filter on closed class word ratio We found that having a low closed class word ratio in a document was one of the best indicators of a non-human generated content. We built lists of closed class words for each language by taking pre-existing lists, for example from Universal Dependencies$^{17}$, which were then reviewed by native speakers. We discard documents with a too low closed class word ratio." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on flagged word ratio To limit the over-representation of pornographic documents, which are in practice much more likely to have shocking and sexist content, and to contain only buzzwords for SEO, we built lists of flagged words for each language by gathering existing lists, and filtering them by native speakers with precise instructions. We are then able to compute the flagged word ratio of a document and discard it if it is too high. About 1% of the documents for each language are removed by this filter.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 452.9870910644531, "coord_origin": "BOTTOMLEFT", "l": 107.08467102050781, "r": 504.1150207519531, "t": 517.4395751953125 }, "charspan": [ 0, 518 ], "page_no": 27 } ], "self_ref": "#/texts/318", "text": "Filter on flagged word ratio To limit the over-representation of pornographic documents, which are in practice much more likely to have shocking and sexist content, and to contain only buzzwords for SEO, we built lists of flagged words for each language by gathering existing lists, and filtering them by native speakers with precise instructions. We are then able to compute the flagged word ratio of a document and discard it if it is too high. About 1% of the documents for each language are removed by this filter." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Instructions for building the lists of flagged words: Keep only the words associated with porn and systematically used in a sexual context. Remove words that can be used in medical, scientific, colloquial (without referring systematically to porn), or everyday contexts. Remove all insults. Remove all words referring to race or sexual orientation.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 403.27685546875, "coord_origin": "BOTTOMLEFT", "l": 107.07425689697266, "r": 505.7449645996094, "t": 446.03607177734375 }, "charspan": [ 0, 348 ], "page_no": 27 } ], "self_ref": "#/texts/319", "text": "Instructions for building the lists of flagged words: Keep only the words associated with porn and systematically used in a sexual context. Remove words that can be used in medical, scientific, colloquial (without referring systematically to porn), or everyday contexts. Remove all insults. Remove all words referring to race or sexual orientation." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on language identification prediction score We used fastText (Joulin et al., 2017) to perform language identification and getting confidence scores for each document. If a score is below a specific threshold, we discard the document. We chose to eliminate few documents with this filter, because the language identification does not perform as well on low-resource languages.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 347.809326171875, "coord_origin": "BOTTOMLEFT", "l": 107.31061553955078, "r": 505.6579284667969, "t": 390.3705749511719 }, "charspan": [ 0, 382 ], "page_no": 27 } ], "self_ref": "#/texts/320", "text": "Filter on language identification prediction score We used fastText (Joulin et al., 2017) to perform language identification and getting confidence scores for each document. If a score is below a specific threshold, we discard the document. We chose to eliminate few documents with this filter, because the language identification does not perform as well on low-resource languages." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Filter on perplexity score Following Wenzek et al. (2020), we trained SentencePiece unigram tokenizers (Kudo, 2018) followed by KenLM 5-gram models after tokenization (Heafield, 2011) on Wikipedia article openings for every language that was extracted from OSCAR. As in De la Rosa et al. (2022), we discarded documents to move the perplexity distribution towards the median, to avoid too high perplexity scores (deemed as not useful for the model), but subsampling was done by perplexity thresholding, not by reshaping the distribution as in De la Rosa et al. (2022). This thresholding was done lightly, by having native speakers manually establish the cutoff values per language$^{18}$, so as not to be too biased by the Wikipedia content and keep the dataset diverse.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 248.38189697265625, "coord_origin": "BOTTOMLEFT", "l": 107.238037109375, "r": 504.1700439453125, "t": 334.35858154296875 }, "charspan": [ 0, 769 ], "page_no": 27 } ], "self_ref": "#/texts/321", "text": "Filter on perplexity score Following Wenzek et al. (2020), we trained SentencePiece unigram tokenizers (Kudo, 2018) followed by KenLM 5-gram models after tokenization (Heafield, 2011) on Wikipedia article openings for every language that was extracted from OSCAR. As in De la Rosa et al. (2022), we discarded documents to move the perplexity distribution towards the median, to avoid too high perplexity scores (deemed as not useful for the model), but subsampling was done by perplexity thresholding, not by reshaping the distribution as in De la Rosa et al. (2022). This thresholding was done lightly, by having native speakers manually establish the cutoff values per language$^{18}$, so as not to be too biased by the Wikipedia content and keep the dataset diverse." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "D PII filtering initiative", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 219.25994873046875, "coord_origin": "BOTTOMLEFT", "l": 107.71265411376953, "r": 236.61404418945312, "t": 230.80224609375 }, "charspan": [ 0, 26 ], "page_no": 27 } ], "self_ref": "#/texts/322", "text": "D PII filtering initiative" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Even if not eventually used in our final pipeline, we have released muliwai 19 a library for text preprocessing, augmentation, anonymization, and synthesis. It relies on transformer models and backtranslation to perform NER and associated augmentation and anonymization over 100+ languages (i.e., we rely on XLMRoberta Fan et al. (2021) and M2M100 Conneau et al. (2020)). We either use a specific model for the chosen language or a model with cross-lingual capabilities. Muliwai tags using the aforementioned transformer then translate the sentence to a target language (e.g., English) and test to see if the translation preserves the NER tagging and discounts or increases the weight of a NER decision accordingly. It then performs NER in the target language and back translates to", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 119.8280029296875, "coord_origin": "BOTTOMLEFT", "l": 107.10517120361328, "r": 505.658203125, "t": 207.41725158691406 }, "charspan": [ 0, 782 ], "page_no": 27 } ], "self_ref": "#/texts/323", "text": "Even if not eventually used in our final pipeline, we have released muliwai 19 a library for text preprocessing, augmentation, anonymization, and synthesis. It relies on transformer models and backtranslation to perform NER and associated augmentation and anonymization over 100+ languages (i.e., we rely on XLMRoberta Fan et al. (2021) and M2M100 Conneau et al. (2020)). We either use a specific model for the chosen language or a model with cross-lingual capabilities. Muliwai tags using the aforementioned transformer then translate the sentence to a target language (e.g., English) and test to see if the translation preserves the NER tagging and discounts or increases the weight of a NER decision accordingly. It then performs NER in the target language and back translates to" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "$^{17}$https://universaldependencies.org/ $^{18}$Native speakers used an ad-hoc visualization tool built for the https://huggingface.co/spaces/huggingface/text-data-filtering $^{19}$Pronounced \"mu-lee-why\" , Hawaiian for river. https://github.com/ontocord/muliwai/tree/main", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 70.0632553100586, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 462.70098876953125, "t": 109.74321746826172 }, "charspan": [ 0, 273 ], "page_no": 27 } ], "self_ref": "#/texts/324", "text": "$^{17}$https://universaldependencies.org/ $^{18}$Native speakers used an ad-hoc visualization tool built for the https://huggingface.co/spaces/huggingface/text-data-filtering $^{19}$Pronounced \"mu-lee-why\" , Hawaiian for river. https://github.com/ontocord/muliwai/tree/main" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "27", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.37042236328125, "r": 310.9815673828125, "t": 49.9281005859375 }, "charspan": [ 0, 2 ], "page_no": 27 } ], "self_ref": "#/texts/325", "text": "27" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "occasion:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 90.87625885009766, "coord_origin": "BOTTOMLEFT", "l": 470.6994323730469, "r": 505.24285888671875, "t": 98.89221954345703 }, "charspan": [ 0, 9 ], "page_no": 27 } ], "self_ref": "#/texts/326", "text": "occasion:" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "the source language. Finally it matches the translated sentence to the original sentence to determine which text spans in the source language sentence should be NER tagged based on the target language NER. We also use spacy and regex as added signals for NER tags.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 685.7234497070312, "coord_origin": "BOTTOMLEFT", "l": 106.87429809570312, "r": 504.0005798339844, "t": 717.525146484375 }, "charspan": [ 0, 264 ], "page_no": 28 } ], "self_ref": "#/texts/327", "text": "the source language. Finally it matches the translated sentence to the original sentence to determine which text spans in the source language sentence should be NER tagged based on the target language NER. We also use spacy and regex as added signals for NER tags." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We also include in the library specific regexes for detecting age, email, date, time, personal addresses, phone numbers and government-issued identifiers (such as license plates). Some regex matches use also the surrounding text context to improve precision.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 647.549560546875, "coord_origin": "BOTTOMLEFT", "l": 106.99664306640625, "r": 505.241943359375, "t": 679.465576171875 }, "charspan": [ 0, 258 ], "page_no": 28 } ], "self_ref": "#/texts/328", "text": "We also include in the library specific regexes for detecting age, email, date, time, personal addresses, phone numbers and government-issued identifiers (such as license plates). Some regex matches use also the surrounding text context to improve precision." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "However, the scale of the data, the fact that the impact on the resulting text could not be fully assessed in terms of language modeling and the time constraint due to compute allocation, meant this approach could not be operationalized on ROOTS. Instead we fell back to a simpler approach, see Section 3.3.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 609.6541137695312, "coord_origin": "BOTTOMLEFT", "l": 107.32073974609375, "r": 504.00128173828125, "t": 640.7891235351562 }, "charspan": [ 0, 307 ], "page_no": 28 } ], "self_ref": "#/texts/329", "text": "However, the scale of the data, the fact that the impact on the resulting text could not be fully assessed in terms of language modeling and the time constraint due to compute allocation, meant this approach could not be operationalized on ROOTS. Instead we fell back to a simpler approach, see Section 3.3." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "E Data Sources", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 580.913330078125, "coord_origin": "BOTTOMLEFT", "l": 107.76239013671875, "r": 195.11753845214844, "t": 592.410888671875 }, "charspan": [ 0, 14 ], "page_no": 28 } ], "self_ref": "#/texts/330", "text": "E Data Sources" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "28", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.52325439453125, "r": 310.9817810058594, "t": 49.5931396484375 }, "charspan": [ 0, 2 ], "page_no": 28 } ], "self_ref": "#/texts/331", "text": "28" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "29", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.47039794921875, "r": 311.015380859375, "t": 49.74163818359375 }, "charspan": [ 0, 2 ], "page_no": 29 } ], "self_ref": "#/texts/332", "text": "29" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "30", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.6112365722656, "r": 311.2100830078125, "t": 49.60919189453125 }, "charspan": [ 0, 2 ], "page_no": 30 } ], "self_ref": "#/texts/333", "text": "30" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Dataset", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 121.41758728027344, "r": 153.55062866210938, "t": 717.1884155273438 }, "charspan": [ 0, 7 ], "page_no": 31 } ], "self_ref": "#/texts/334", "text": "Dataset" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Language", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.7261352539062, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 289.263916015625, "t": 716.8920288085938 }, "charspan": [ 0, 8 ], "page_no": 31 } ], "self_ref": "#/texts/335", "text": "Language" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Source", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 371.35443115234375, "r": 401.1227111816406, "t": 716.8472290039062 }, "charspan": [ 0, 6 ], "page_no": 31 } ], "self_ref": "#/texts/336", "text": "Source" }, { "children": [], "enumerated": null, "label": "paragraph", "level": null, "marker": null, "orig": "Table 2: List of datasets used in crowdsourced dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 682.9140625, "coord_origin": "BOTTOMLEFT", "l": 195.41920471191406, "r": 413.2532958984375, "t": 693.0294189453125 }, "charspan": [ 0, 55 ], "page_no": 31 } ], "self_ref": "#/texts/337", "text": "Table 2: List of datasets used in crowdsourced dataset." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "31", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.5538024902344, "r": 310.9815979003906, "t": 49.55767822265625 }, "charspan": [ 0, 2 ], "page_no": 31 } ], "self_ref": "#/texts/338", "text": "31" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Table 3: Linguistic makeup of the corpus.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 141.5594482421875, "coord_origin": "BOTTOMLEFT", "l": 221.92112731933594, "r": 389.23443603515625, "t": 151.5220947265625 }, "charspan": [ 0, 41 ], "page_no": 32 } ], "self_ref": "#/texts/339", "text": "Table 3: Linguistic makeup of the corpus." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "32", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.66900634765625, "r": 311.05078125, "t": 49.68975830078125 }, "charspan": [ 0, 2 ], "page_no": 32 } ], "self_ref": "#/texts/340", "text": "32" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Table 4: Pseudocrawled data per language sorted by number of domains crawled", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 292.16748046875, "coord_origin": "BOTTOMLEFT", "l": 144.45069885253906, "r": 466.5780334472656, "t": 302.23150634765625 }, "charspan": [ 0, 76 ], "page_no": 33 } ], "self_ref": "#/texts/341", "text": "Table 4: Pseudocrawled data per language sorted by number of domains crawled" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "33", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.9600944519043, "coord_origin": "BOTTOMLEFT", "l": 300.5620422363281, "r": 310.9815673828125, "t": 49.5869140625 }, "charspan": [ 0, 2 ], "page_no": 33 } ], "self_ref": "#/texts/342", "text": "33" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "F Author contributions", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.538330078125, "coord_origin": "BOTTOMLEFT", "l": 107.35459899902344, "r": 235.02398681640625, "t": 718.9667358398438 }, "charspan": [ 0, 22 ], "page_no": 34 } ], "self_ref": "#/texts/343", "text": "F Author contributions" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Author contributions in alphabetical order.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 659.5370483398438, "coord_origin": "BOTTOMLEFT", "l": 106.92214965820312, "r": 277.0649719238281, "t": 669.5479125976562 }, "charspan": [ 0, 43 ], "page_no": 34 } ], "self_ref": "#/texts/344", "text": "Author contributions in alphabetical order." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Aaron Gokaslan set a pre-commit (for code formatting) in a repository and helped with the writing of the Related Work section of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 631.7971801757812, "coord_origin": "BOTTOMLEFT", "l": 106.79485321044922, "r": 504.07818603515625, "t": 652.7407836914062 }, "charspan": [ 0, 139 ], "page_no": 34 } ], "self_ref": "#/texts/345", "text": "Aaron Gokaslan set a pre-commit (for code formatting) in a repository and helped with the writing of the Related Work section of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Aitor Soroa integrated one dataset into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 615.6710205078125, "coord_origin": "BOTTOMLEFT", "l": 106.59163665771484, "r": 344.45269775390625, "t": 625.4960327148438 }, "charspan": [ 0, 58 ], "page_no": 34 } ], "self_ref": "#/texts/346", "text": "Aitor Soroa integrated one dataset into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Albert Villanova del Moral led the gathering of identified sources, implemented loading scripts in the datasets library in a single unified interface, and integrated the most datasets into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 577.6441040039062, "coord_origin": "BOTTOMLEFT", "l": 107.04678344726562, "r": 504.00421142578125, "t": 608.8527221679688 }, "charspan": [ 0, 207 ], "page_no": 34 } ], "self_ref": "#/texts/347", "text": "Albert Villanova del Moral led the gathering of identified sources, implemented loading scripts in the datasets library in a single unified interface, and integrated the most datasets into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Angelina McMillan-Major gathered the lists of closed class words for many languages used for the filtering of OSCAR.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 550.3460693359375, "coord_origin": "BOTTOMLEFT", "l": 106.93609619140625, "r": 503.9969482421875, "t": 571.01123046875 }, "charspan": [ 0, 116 ], "page_no": 34 } ], "self_ref": "#/texts/348", "text": "Angelina McMillan-Major gathered the lists of closed class words for many languages used for the filtering of OSCAR." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Anna Rogers contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 533.371826171875, "coord_origin": "BOTTOMLEFT", "l": 106.6614761352539, "r": 319.1272277832031, "t": 543.545654296875 }, "charspan": [ 0, 52 ], "page_no": 34 } ], "self_ref": "#/texts/349", "text": "Anna Rogers contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Chenghao Mou was the main contributor for OSCAR deduplication.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 517.569091796875, "coord_origin": "BOTTOMLEFT", "l": 107.25872039794922, "r": 384.15289306640625, "t": 527.6318359375 }, "charspan": [ 0, 62 ], "page_no": 34 } ], "self_ref": "#/texts/350", "text": "Chenghao Mou was the main contributor for OSCAR deduplication." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Christopher Akiki advised on analysis aspects of the project, integrated over a hundred datasets into crowdsourced data, wrote dataset loading scripts, participated in cleaning and filtering efforts, helped with visualization, and contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 479.135009765625, "coord_origin": "BOTTOMLEFT", "l": 107.0223617553711, "r": 504.06256103515625, "t": 510.9559631347656 }, "charspan": [ 0, 271 ], "page_no": 34 } ], "self_ref": "#/texts/351", "text": "Christopher Akiki advised on analysis aspects of the project, integrated over a hundred datasets into crowdsourced data, wrote dataset loading scripts, participated in cleaning and filtering efforts, helped with visualization, and contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Daniel van Strien integrated one dataset into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 462.9740905761719, "coord_origin": "BOTTOMLEFT", "l": 107.50894165039062, "r": 368.9797058105469, "t": 472.83099365234375 }, "charspan": [ 0, 64 ], "page_no": 34 } ], "self_ref": "#/texts/352", "text": "Daniel van Strien integrated one dataset into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "David Ifeoluwa Adelani participated in the PII filtering initiative (see Appendix D).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 446.3478088378906, "coord_origin": "BOTTOMLEFT", "l": 107.58077239990234, "r": 444.9547119140625, "t": 456.3094177246094 }, "charspan": [ 0, 85 ], "page_no": 34 } ], "self_ref": "#/texts/353", "text": "David Ifeoluwa Adelani participated in the PII filtering initiative (see Appendix D)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Eduardo González Ponferrada trained SentencePiece and KenLM models used for the filtering of OSCAR.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 419.2880859375, "coord_origin": "BOTTOMLEFT", "l": 107.3494644165039, "r": 504.7056884765625, "t": 440.16180419921875 }, "charspan": [ 0, 99 ], "page_no": 34 } ], "self_ref": "#/texts/354", "text": "Eduardo González Ponferrada trained SentencePiece and KenLM models used for the filtering of OSCAR." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Francesco De Toni led one crowdsourcing hackathon, analyzed the distribution of the sources in the catalogue, participated in the PII filtering initiative (see Appendix D), and contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 380.6253356933594, "coord_origin": "BOTTOMLEFT", "l": 107.1496353149414, "r": 504.0037536621094, "t": 412.8728332519531 }, "charspan": [ 0, 217 ], "page_no": 34 } ], "self_ref": "#/texts/355", "text": "Francesco De Toni led one crowdsourcing hackathon, analyzed the distribution of the sources in the catalogue, participated in the PII filtering initiative (see Appendix D), and contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Giada Pistilli helped write the Ethical Considerations section of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 363.9192810058594, "coord_origin": "BOTTOMLEFT", "l": 107.5256576538086, "r": 406.03045654296875, "t": 374.1573181152344 }, "charspan": [ 0, 76 ], "page_no": 34 } ], "self_ref": "#/texts/356", "text": "Giada Pistilli helped write the Ethical Considerations section of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Gérard M Dupont contributed to data tooling and sourcing and advised on analysis aspects of the project.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 336.9927673339844, "coord_origin": "BOTTOMLEFT", "l": 107.22802734375, "r": 503.99615478515625, "t": 358.14361572265625 }, "charspan": [ 0, 104 ], "page_no": 34 } ], "self_ref": "#/texts/357", "text": "Gérard M Dupont contributed to data tooling and sourcing and advised on analysis aspects of the project." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hieu Tran helped set the filtering parameters for Vietnamese and contributed to the list of Vietnamese closed-class words.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 310.0970764160156, "coord_origin": "BOTTOMLEFT", "l": 107.21896362304688, "r": 503.9962463378906, "t": 330.9805603027344 }, "charspan": [ 0, 122 ], "page_no": 34 } ], "self_ref": "#/texts/358", "text": "Hieu Tran helped set the filtering parameters for Vietnamese and contributed to the list of Vietnamese closed-class words." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hugo Laurençon developed the filtering library used for the cleaning of OSCAR and the visualization tool to help choose the filtering parameters, and ran OSCAR filtering jobs. He was involved in the cleaning of some crowdsourced and pseudo-crawled datasets and the deduplication of OSCAR. He also contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 260.76324462890625, "coord_origin": "BOTTOMLEFT", "l": 107.11961364746094, "r": 504.0040283203125, "t": 303.48046875 }, "charspan": [ 0, 337 ], "page_no": 34 } ], "self_ref": "#/texts/359", "text": "Hugo Laurençon developed the filtering library used for the cleaning of OSCAR and the visualization tool to help choose the filtering parameters, and ran OSCAR filtering jobs. He was involved in the cleaning of some crowdsourced and pseudo-crawled datasets and the deduplication of OSCAR. He also contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Huu Nguyen contributed to the data tooling and lead the PII filtering initiative (see Appendix D).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 244.24407958984375, "coord_origin": "BOTTOMLEFT", "l": 107.7374267578125, "r": 497.29931640625, "t": 254.33538818359375 }, "charspan": [ 0, 98 ], "page_no": 34 } ], "self_ref": "#/texts/360", "text": "Huu Nguyen contributed to the data tooling and lead the PII filtering initiative (see Appendix D)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ian Yu participated in the PII filtering initiative (see Appendix D) and helped to choose the filtering parameters for Chinese.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 217.2950897216797, "coord_origin": "BOTTOMLEFT", "l": 107.30577850341797, "r": 503.997314453125, "t": 237.974365234375 }, "charspan": [ 0, 127 ], "page_no": 34 } ], "self_ref": "#/texts/361", "text": "Ian Yu participated in the PII filtering initiative (see Appendix D) and helped to choose the filtering parameters for Chinese." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Itziar Gonzalez-Dios , as a Basque native speaker, helped choose the filtering parameters for this language.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 189.27911376953125, "coord_origin": "BOTTOMLEFT", "l": 107.51122283935547, "r": 504.12860107421875, "t": 210.8109130859375 }, "charspan": [ 0, 108 ], "page_no": 34 } ], "self_ref": "#/texts/362", "text": "Itziar Gonzalez-Dios , as a Basque native speaker, helped choose the filtering parameters for this language." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Javier De la Rosa contributed with perplexity sampling efforts for OSCAR (not used in final pipeline).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 162.09222412109375, "coord_origin": "BOTTOMLEFT", "l": 106.78717041015625, "r": 504.0043640136719, "t": 183.446044921875 }, "charspan": [ 0, 102 ], "page_no": 34 } ], "self_ref": "#/texts/363", "text": "Javier De la Rosa contributed with perplexity sampling efforts for OSCAR (not used in final pipeline)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jenny Chim participated in the PII filtering initiative (see Appendix D) and helped to choose the filtering parameters and closed-class words for Chinese.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 135.03594970703125, "coord_origin": "BOTTOMLEFT", "l": 106.87274932861328, "r": 503.9988708496094, "t": 156.2374267578125 }, "charspan": [ 0, 154 ], "page_no": 34 } ], "self_ref": "#/texts/364", "text": "Jenny Chim participated in the PII filtering initiative (see Appendix D) and helped to choose the filtering parameters and closed-class words for Chinese." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jian Zhu integrated two datasets into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 119.01407623291016, "coord_origin": "BOTTOMLEFT", "l": 106.80430603027344, "r": 337.4784240722656, "t": 128.9278564453125 }, "charspan": [ 0, 56 ], "page_no": 34 } ], "self_ref": "#/texts/365", "text": "Jian Zhu integrated two datasets into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jörg Frohberg integrated multiple datasets into crowdsourced data and reached out to license holders.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 102.1832275390625, "coord_origin": "BOTTOMLEFT", "l": 106.791748046875, "r": 505.74700927734375, "t": 112.29534912109375 }, "charspan": [ 0, 101 ], "page_no": 34 } ], "self_ref": "#/texts/366", "text": "Jörg Frohberg integrated multiple datasets into crowdsourced data and reached out to license holders." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Khalid Almubarak integrated some datasets into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 86.23707580566406, "coord_origin": "BOTTOMLEFT", "l": 107.55047607421875, "r": 386.53424072265625, "t": 96.46240234375 }, "charspan": [ 0, 65 ], "page_no": 34 } ], "self_ref": "#/texts/367", "text": "Khalid Almubarak integrated some datasets into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kyle Lo integrated one dataset into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.60845947265625, "coord_origin": "BOTTOMLEFT", "l": 107.81101989746094, "r": 328.501708984375, "t": 79.48779296875 }, "charspan": [ 0, 54 ], "page_no": 34 } ], "self_ref": "#/texts/368", "text": "Kyle Lo integrated one dataset into crowdsourced data." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "34", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.4390563964844, "r": 310.9815979003906, "t": 49.51678466796875 }, "charspan": [ 0, 2 ], "page_no": 34 } ], "self_ref": "#/texts/369", "text": "34" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Leandro von Werra participated in cleaning and filtering efforts, built the code dataset, contributed to its analysis, and participated in the sourcing effort for target datasets.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.7778930664062, "coord_origin": "BOTTOMLEFT", "l": 107.24435424804688, "r": 504.0033264160156, "t": 717.5348510742188 }, "charspan": [ 0, 179 ], "page_no": 35 } ], "self_ref": "#/texts/370", "text": "Leandro von Werra participated in cleaning and filtering efforts, built the code dataset, contributed to its analysis, and participated in the sourcing effort for target datasets." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Leon Weber integrated one dataset into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 680.4395141601562, "coord_origin": "BOTTOMLEFT", "l": 107.4734115600586, "r": 347.1127014160156, "t": 690.1265869140625 }, "charspan": [ 0, 57 ], "page_no": 35 } ], "self_ref": "#/texts/371", "text": "Leon Weber integrated one dataset into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Long Phan participated in the PII filtering initiative (see Appendix D).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 664.1199340820312, "coord_origin": "BOTTOMLEFT", "l": 107.3431396484375, "r": 391.2467041015625, "t": 674.2333984375 }, "charspan": [ 0, 72 ], "page_no": 35 } ], "self_ref": "#/texts/372", "text": "Long Phan participated in the PII filtering initiative (see Appendix D)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Loubna Ben allal contributed the analysis of the code dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 647.5546875, "coord_origin": "BOTTOMLEFT", "l": 107.32406616210938, "r": 357.4134521484375, "t": 657.7589111328125 }, "charspan": [ 0, 62 ], "page_no": 35 } ], "self_ref": "#/texts/373", "text": "Loubna Ben allal contributed the analysis of the code dataset." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Lucile Saulnier co-led pseudo-crawled data acquisition, contributed to filtering, cleaning, and deduplication for the crowdsourced datasets, built visualization tools to inspect the results of preprocessing, scaled the PII filtering process, performed the document size analysis, and participated in paper writing.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 597.7581787109375, "coord_origin": "BOTTOMLEFT", "l": 107.2283706665039, "r": 505.65362548828125, "t": 640.9046630859375 }, "charspan": [ 0, 314 ], "page_no": 35 } ], "self_ref": "#/texts/374", "text": "Lucile Saulnier co-led pseudo-crawled data acquisition, contributed to filtering, cleaning, and deduplication for the crowdsourced datasets, built visualization tools to inspect the results of preprocessing, scaled the PII filtering process, performed the document size analysis, and participated in paper writing." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Manan Dey , as a Bengali native speaker, helped to choose the filtering parameters for this language.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 582.1322631835938, "coord_origin": "BOTTOMLEFT", "l": 107.349609375, "r": 505.6278991699219, "t": 592.245361328125 }, "charspan": [ 0, 101 ], "page_no": 35 } ], "self_ref": "#/texts/375", "text": "Manan Dey , as a Bengali native speaker, helped to choose the filtering parameters for this language." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Manuel Romero Muñoz contributed to KenLM models and to corpus visualization with those models, and participated in the PII filtering initiative (see Appendix D).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 554.859130859375, "coord_origin": "BOTTOMLEFT", "l": 107.21385192871094, "r": 504.0985107421875, "t": 575.7548217773438 }, "charspan": [ 0, 161 ], "page_no": 35 } ], "self_ref": "#/texts/376", "text": "Manuel Romero Muñoz contributed to KenLM models and to corpus visualization with those models, and participated in the PII filtering initiative (see Appendix D)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Maraim Masoud contributed to the sourcing of some Arabic datasets, the list of Arabic closed-class words, and the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 527.2988891601562, "coord_origin": "BOTTOMLEFT", "l": 107.14813995361328, "r": 504.10321044921875, "t": 548.3705444335938 }, "charspan": [ 0, 135 ], "page_no": 35 } ], "self_ref": "#/texts/377", "text": "Maraim Masoud contributed to the sourcing of some Arabic datasets, the list of Arabic closed-class words, and the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Margaret Mitchell co-led the final regex-based PII efforts.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 511.15283203125, "coord_origin": "BOTTOMLEFT", "l": 107.24675750732422, "r": 344.5412902832031, "t": 521.038330078125 }, "charspan": [ 0, 59 ], "page_no": 35 } ], "self_ref": "#/texts/378", "text": "Margaret Mitchell co-led the final regex-based PII efforts." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Mario Šaško integrated multiple datasets into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 493.55108642578125, "coord_origin": "BOTTOMLEFT", "l": 107.25186157226562, "r": 371.4304504394531, "t": 504.02252197265625 }, "charspan": [ 0, 64 ], "page_no": 35 } ], "self_ref": "#/texts/379", "text": "Mario Šaško integrated multiple datasets into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Olivier Nguyen helped build the first blocks of the OSCAR filtering pipeline.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 476.7587585449219, "coord_origin": "BOTTOMLEFT", "l": 107.40467071533203, "r": 418.10565185546875, "t": 487.0457458496094 }, "charspan": [ 0, 77 ], "page_no": 35 } ], "self_ref": "#/texts/380", "text": "Olivier Nguyen helped build the first blocks of the OSCAR filtering pipeline." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Paulo Villegas participated in the PII filtering initiative (see Appendix D) and the perplexity sampling efforts for OSCAR.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 450.08209228515625, "coord_origin": "BOTTOMLEFT", "l": 107.23342895507812, "r": 503.9970397949219, "t": 470.8218688964844 }, "charspan": [ 0, 123 ], "page_no": 35 } ], "self_ref": "#/texts/381", "text": "Paulo Villegas participated in the PII filtering initiative (see Appendix D) and the perplexity sampling efforts for OSCAR." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Pedro Ortiz Suarez contributed to crowdsourced data and provided high-level metrics on OSCAR.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 433.6725158691406, "coord_origin": "BOTTOMLEFT", "l": 107.3623046875, "r": 504.6109619140625, "t": 443.6438903808594 }, "charspan": [ 0, 93 ], "page_no": 35 } ], "self_ref": "#/texts/382", "text": "Pedro Ortiz Suarez contributed to crowdsourced data and provided high-level metrics on OSCAR." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Pierre Colombo participated in the PII filtering initiative (see Appendix D).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 416.7821350097656, "coord_origin": "BOTTOMLEFT", "l": 107.33390808105469, "r": 412.06866455078125, "t": 427.1553955078125 }, "charspan": [ 0, 77 ], "page_no": 35 } ], "self_ref": "#/texts/383", "text": "Pierre Colombo participated in the PII filtering initiative (see Appendix D)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Quentin Lhoest integrated multiple datasets into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 400.9170837402344, "coord_origin": "BOTTOMLEFT", "l": 107.47139739990234, "r": 383.7644348144531, "t": 411.0438232421875 }, "charspan": [ 0, 67 ], "page_no": 35 } ], "self_ref": "#/texts/384", "text": "Quentin Lhoest integrated multiple datasets into crowdsourced data." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Sasha Luccioni co-led the final regex-based PII efforts. participated in filtering and cleaning efforts, and contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 373.3299255371094, "coord_origin": "BOTTOMLEFT", "l": 107.42141723632812, "r": 505.38671875, "t": 394.1334533691406 }, "charspan": [ 0, 149 ], "page_no": 35 } ], "self_ref": "#/texts/385", "text": "Sasha Luccioni co-led the final regex-based PII efforts. participated in filtering and cleaning efforts, and contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Sebastian Nagel helped to implement the pseudo-crawled data acquisition step.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 356.77703857421875, "coord_origin": "BOTTOMLEFT", "l": 107.16508483886719, "r": 426.623779296875, "t": 366.7722473144531 }, "charspan": [ 0, 77 ], "page_no": 35 } ], "self_ref": "#/texts/386", "text": "Sebastian Nagel helped to implement the pseudo-crawled data acquisition step." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Shamik Bose participated in the PII filtering initiative (see Appendix D) and contributed the list of Bengali closed-class words.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 329.6058349609375, "coord_origin": "BOTTOMLEFT", "l": 107.36302947998047, "r": 504.3883972167969, "t": 350.10711669921875 }, "charspan": [ 0, 129 ], "page_no": 35 } ], "self_ref": "#/texts/387", "text": "Shamik Bose participated in the PII filtering initiative (see Appendix D) and contributed the list of Bengali closed-class words." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Shayne Longpre contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 313.34368896484375, "coord_origin": "BOTTOMLEFT", "l": 107.45349884033203, "r": 333.7122497558594, "t": 323.61883544921875 }, "charspan": [ 0, 55 ], "page_no": 35 } ], "self_ref": "#/texts/388", "text": "Shayne Longpre contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Somaieh Nikpoor co-led the development of ethical/legal charter and contributed to the section on ethical considerations.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 286.2470703125, "coord_origin": "BOTTOMLEFT", "l": 107.18673706054688, "r": 504.12335205078125, "t": 307.0542297363281 }, "charspan": [ 0, 121 ], "page_no": 35 } ], "self_ref": "#/texts/389", "text": "Somaieh Nikpoor co-led the development of ethical/legal charter and contributed to the section on ethical considerations." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Stella Biderman contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 269.41729736328125, "coord_origin": "BOTTOMLEFT", "l": 107.3687973022461, "r": 332.7752380371094, "t": 279.8677978515625 }, "charspan": [ 0, 56 ], "page_no": 35 } ], "self_ref": "#/texts/390", "text": "Stella Biderman contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Suhas Pai participated in the PII filtering initiative (see Appendix D).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 252.97271728515625, "coord_origin": "BOTTOMLEFT", "l": 107.24922180175781, "r": 386.1656799316406, "t": 262.90374755859375 }, "charspan": [ 0, 72 ], "page_no": 35 } ], "self_ref": "#/texts/391", "text": "Suhas Pai participated in the PII filtering initiative (see Appendix D)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Suzana Ili'c coordinated the organization of BigScience working groups.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 236.8934326171875, "coord_origin": "BOTTOMLEFT", "l": 107.29518127441406, "r": 397.9416198730469, "t": 246.9400634765625 }, "charspan": [ 0, 71 ], "page_no": 35 } ], "self_ref": "#/texts/392", "text": "Suzana Ili'c coordinated the organization of BigScience working groups." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Teven Le Scao led final quality control checks, contributed to filtering, cleaning, and deduplication for all components of the corpus, contributed to the OSCAR filtering visualization tool, contributed and repaired several datasets for crowdsourced data, performed the tokenizer-based analysis, and participated in the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 187.24859619140625, "coord_origin": "BOTTOMLEFT", "l": 107.14020538330078, "r": 504.0654602050781, "t": 230.31463623046875 }, "charspan": [ 0, 341 ], "page_no": 35 } ], "self_ref": "#/texts/393", "text": "Teven Le Scao led final quality control checks, contributed to filtering, cleaning, and deduplication for all components of the corpus, contributed to the OSCAR filtering visualization tool, contributed and repaired several datasets for crowdsourced data, performed the tokenizer-based analysis, and participated in the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Thomas Wang co-led pseudo-crawled data acquisition, built the distributed cleaning pipelines for pseudo-crawled and crowdsourced datasets, handled job monitoring for crowdsourced dataset filtering, and participated in paper writing.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 149.19110107421875, "coord_origin": "BOTTOMLEFT", "l": 106.96576690673828, "r": 505.2413024902344, "t": 181.48101806640625 }, "charspan": [ 0, 232 ], "page_no": 35 } ], "self_ref": "#/texts/394", "text": "Thomas Wang co-led pseudo-crawled data acquisition, built the distributed cleaning pipelines for pseudo-crawled and crowdsourced datasets, handled job monitoring for crowdsourced dataset filtering, and participated in paper writing." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Tristan Thrush participated in the final regex-based PII efforts.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 132.78717041015625, "coord_origin": "BOTTOMLEFT", "l": 106.88078308105469, "r": 362.93646240234375, "t": 142.8299560546875 }, "charspan": [ 0, 65 ], "page_no": 35 } ], "self_ref": "#/texts/395", "text": "Tristan Thrush participated in the final regex-based PII efforts." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Violette Lepercq was the primary project manager for the final dataset cleaning efforts and helped to reach out to the native speakers to tune the filtering parameters.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 105.3494873046875, "coord_origin": "BOTTOMLEFT", "l": 107.00873565673828, "r": 504.09814453125, "t": 126.7177734375 }, "charspan": [ 0, 168 ], "page_no": 35 } ], "self_ref": "#/texts/396", "text": "Violette Lepercq was the primary project manager for the final dataset cleaning efforts and helped to reach out to the native speakers to tune the filtering parameters." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Vu Minh Chien participated in the PII filtering initiative (see Appendix D).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 89.45928955078125, "coord_origin": "BOTTOMLEFT", "l": 107.01558685302734, "r": 409.8509216308594, "t": 99.57073974609375 }, "charspan": [ 0, 76 ], "page_no": 35 } ], "self_ref": "#/texts/397", "text": "Vu Minh Chien participated in the PII filtering initiative (see Appendix D)." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "35", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.5043640136719, "r": 310.9815673828125, "t": 49.5968017578125 }, "charspan": [ 0, 2 ], "page_no": 35 } ], "self_ref": "#/texts/398", "text": "35" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Yacine Jernite defined the primary goals of the project, advised on data collection and filtering efforts, contributed sourcing tools and early cleaning scripts, and contributed to the writing of the paper.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 685.6644287109375, "coord_origin": "BOTTOMLEFT", "l": 106.8611831665039, "r": 504.00390625, "t": 717.7649536132812 }, "charspan": [ 0, 206 ], "page_no": 36 } ], "self_ref": "#/texts/399", "text": "Yacine Jernite defined the primary goals of the project, advised on data collection and filtering efforts, contributed sourcing tools and early cleaning scripts, and contributed to the writing of the paper." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Zaid Alyafeai helped find a list of Arabic datasets that should be integrated into crowdsourced data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 669.4737548828125, "coord_origin": "BOTTOMLEFT", "l": 107.1594467163086, "r": 500.96099853515625, "t": 679.447265625 }, "charspan": [ 0, 101 ], "page_no": 36 } ], "self_ref": "#/texts/400", "text": "Zaid Alyafeai helped find a list of Arabic datasets that should be integrated into crowdsourced data." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "36", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.6223449707031, "r": 311.20245361328125, "t": 49.731201171875 }, "charspan": [ 0, 2 ], "page_no": 36 } ], "self_ref": "#/texts/401", "text": "36" } ]

[ { "annotations": [], "captions": [], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 567.3838500976562, "coord_origin": "BOTTOMLEFT", "l": 108.22891235351562, "r": 384.4786376953125, "t": 719.500244140625 }, "charspan": [ 0, 0 ], "page_no": 3 } ], "references": [], "self_ref": "#/pictures/0" }, { "annotations": [], "captions": [ { "$ref": "#/texts/17" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 567.1375732421875, "coord_origin": "BOTTOMLEFT", "l": 419.9916687011719, "r": 503.9209899902344, "t": 721.4024658203125 }, "charspan": [ 0, 519 ], "page_no": 3 } ], "references": [], "self_ref": "#/pictures/1" }, { "annotations": [], "captions": [ { "$ref": "#/texts/56" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 643.2068481445312, "coord_origin": "BOTTOMLEFT", "l": 108.33333587646484, "r": 502.23834228515625, "t": 718.7099609375 }, "charspan": [ 0, 235 ], "page_no": 7 } ], "references": [], "self_ref": "#/pictures/2" }, { "annotations": [], "captions": [ { "$ref": "#/texts/73" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 583.3694458007812, "coord_origin": "BOTTOMLEFT", "l": 109.66666412353516, "r": 502.16204833984375, "t": 654.992919921875 }, "charspan": [ 0, 88 ], "page_no": 8 } ], "references": [], "self_ref": "#/pictures/3" }, { "annotations": [], "captions": [ { "$ref": "#/texts/84" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 547.3536987304688, "coord_origin": "BOTTOMLEFT", "l": 192.23777770996094, "r": 420.5435485839844, "t": 715.2022705078125 }, "charspan": [ 0, 386 ], "page_no": 9 } ], "references": [], "self_ref": "#/pictures/4" }, { "annotations": [], "captions": [ { "$ref": "#/texts/85" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 366.7183837890625, "coord_origin": "BOTTOMLEFT", "l": 110.748779296875, "r": 501.9324035644531, "t": 475.7723388671875 }, "charspan": [ 0, 233 ], "page_no": 9 } ], "references": [], "self_ref": "#/pictures/5" }, { "annotations": [], "captions": [ { "$ref": "#/texts/94" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 637.3893432617188, "coord_origin": "BOTTOMLEFT", "l": 107.7559585571289, "r": 504.5894775390625, "t": 721.303466796875 }, "charspan": [ 0, 269 ], "page_no": 10 } ], "references": [], "self_ref": "#/pictures/6" }, { "annotations": [], "captions": [ { "$ref": "#/texts/95" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 449.4361572265625, "coord_origin": "BOTTOMLEFT", "l": 108.6402587890625, "r": 503.7072448730469, "t": 585.3333129882812 }, "charspan": [ 0, 248 ], "page_no": 10 } ], "references": [], "self_ref": "#/pictures/7" }, { "annotations": [], "captions": [ { "$ref": "#/texts/102" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 637.7144775390625, "coord_origin": "BOTTOMLEFT", "l": 108.2791748046875, "r": 502.65350341796875, "t": 719.555908203125 }, "charspan": [ 0, 219 ], "page_no": 11 } ], "references": [], "self_ref": "#/pictures/8" }, { "annotations": [], "captions": [ { "$ref": "#/texts/235" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 414.1766357421875, "coord_origin": "BOTTOMLEFT", "l": 135.03651428222656, "r": 483.3057556152344, "t": 511.0146484375 }, "charspan": [ 0, 173 ], "page_no": 22 } ], "references": [], "self_ref": "#/pictures/9" }, { "annotations": [], "captions": [ { "$ref": "#/texts/261" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 604.4828491210938, "coord_origin": "BOTTOMLEFT", "l": 111.65937805175781, "r": 504.4784851074219, "t": 701.9296264648438 }, "charspan": [ 0, 15 ], "page_no": 24 } ], "references": [], "self_ref": "#/pictures/10" }, { "annotations": [], "captions": [ { "$ref": "#/texts/263" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 468.1575927734375, "coord_origin": "BOTTOMLEFT", "l": 110.87045288085938, "r": 505.1271057128906, "t": 564 }, "charspan": [ 0, 15 ], "page_no": 24 } ], "references": [], "self_ref": "#/pictures/11" }, { "annotations": [], "captions": [ { "$ref": "#/texts/264" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 166.732666015625, "coord_origin": "BOTTOMLEFT", "l": 113.39810180664062, "r": 502.1221618652344, "t": 437.8569641113281 }, "charspan": [ 0, 55 ], "page_no": 24 } ], "references": [], "self_ref": "#/pictures/12" } ]

[ { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 681.6038208007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 177.0109405517578, "t": 690.5601806640625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "1 Introduction" }, { "bbox": { "b": 681.6038208007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 177.0109405517578, "t": 690.5601806640625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "1 Introduction" }, { "bbox": { "b": 681.6038208007812, "coord_origin": "BOTTOMLEFT", "l": 499.0220947265625, "r": 504.0033874511719, "t": 690.5601806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "3" } ], [ { "bbox": { "b": 665.1460571289062, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 674.0526123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "1.1" }, { "bbox": { "b": 665.1460571289062, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970764160156, "t": 674.0526123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Outline of the Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 665.1460571289062, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 674.0526123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "3" } ], [ { "bbox": { "b": 648.757080078125, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 657.6636352539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "1.2" }, { "bbox": { "b": 648.757080078125, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970458984375, "t": 657.6636352539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 648.757080078125, "coord_origin": "BOTTOMLEFT", "l": 499.0222473144531, "r": 504.0035400390625, "t": 657.6636352539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "4" } ], [ { "bbox": { "b": 622.475830078125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 112.9813003540039, "t": 631.4321899414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "2" }, { "bbox": { "b": 622.475830078125, "coord_origin": "BOTTOMLEFT", "l": 122.94390106201172, "r": 504.0033874511719, "t": 631.4321899414062 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "(Crowd) Sourcing a Language Resource Catalogue 4" }, { "bbox": { "b": 622.475830078125, "coord_origin": "BOTTOMLEFT", "l": 122.94390106201172, "r": 504.0033874511719, "t": 631.4321899414062 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "(Crowd) Sourcing a Language Resource Catalogue 4" } ], [ { "bbox": { "b": 606.01708984375, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 614.9236450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "2.1" }, { "bbox": { "b": 606.01708984375, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.1070251464844, "t": 614.9236450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "Obtaining Data from the Identified Resources . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 606.01708984375, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 614.9236450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "5" } ], [ { "bbox": { "b": 589.6290893554688, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 598.53564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "2.2" }, { "bbox": { "b": 589.6290893554688, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.10699462890625, "t": 598.53564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Processing Pipeline for Quality Improvement on Crowdsourced Datasets . . . . . ." }, { "bbox": { "b": 589.6290893554688, "coord_origin": "BOTTOMLEFT", "l": 499.022216796875, "r": 504.0035095214844, "t": 598.53564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "6" } ], [ { "bbox": { "b": 563.3478393554688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 572.30419921875 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "3 Processing OSCAR 7" }, { "bbox": { "b": 563.3478393554688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 572.30419921875 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "3 Processing OSCAR 7" }, { "bbox": { "b": 563.3478393554688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 572.30419921875 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "3 Processing OSCAR 7" } ], [ { "bbox": { "b": 546.8890991210938, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 555.795654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "3.1" }, { "bbox": { "b": 546.8890991210938, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.09710693359375, "t": 555.795654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Data cleaning and filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 546.8890991210938, "coord_origin": "BOTTOMLEFT", "l": 499.0223083496094, "r": 504.00360107421875, "t": 555.795654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "7" } ], [ { "bbox": { "b": 530.5010986328125, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 539.4076538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "3.2" }, { "bbox": { "b": 530.5010986328125, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970458984375, "t": 539.4076538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Deduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 530.5010986328125, "coord_origin": "BOTTOMLEFT", "l": 499.0222473144531, "r": 504.0035400390625, "t": 539.4076538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "8" } ], [ { "bbox": { "b": 514.1121215820312, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 523.0186767578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "3.3" }, { "bbox": { "b": 514.1121215820312, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.1070556640625, "t": 523.0186767578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "Personally identifiable information . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 514.1121215820312, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 523.0186767578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "8" } ], [ { "bbox": { "b": 487.8308410644531, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 496.7872314453125 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "4 A First look at ROOTS 8" }, { "bbox": { "b": 487.8308410644531, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 496.7872314453125 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "4 A First look at ROOTS 8" }, { "bbox": { "b": 487.8308410644531, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 496.7872314453125 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "4 A First look at ROOTS 8" } ], [ { "bbox": { "b": 471.3721008300781, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 480.2786560058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "4.1" }, { "bbox": { "b": 471.3721008300781, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970764160156, "t": 480.2786560058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "Natural Languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 471.3721008300781, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 480.2786560058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "8" } ], [ { "bbox": { "b": 454.9841003417969, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 463.8906555175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "4.2" }, { "bbox": { "b": 454.9841003417969, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0971374511719, "t": 463.8906555175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "Programming Languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 454.9841003417969, "coord_origin": "BOTTOMLEFT", "l": 499.0223388671875, "r": 504.0036315917969, "t": 463.8906555175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "9" } ], [ { "bbox": { "b": 438.5950927734375, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 447.50164794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "4.3" }, { "bbox": { "b": 438.5950927734375, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.1070251464844, "t": 447.50164794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "Tokenizer analysis of the component datasets . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 438.5950927734375, "coord_origin": "BOTTOMLEFT", "l": 494.04095458984375, "r": 504.0035400390625, "t": 447.50164794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "10" } ], [ { "bbox": { "b": 412.3138122558594, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 170.55516052246094, "t": 421.27020263671875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "5 Conclusion" }, { "bbox": { "b": 412.3138122558594, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 170.55516052246094, "t": 421.27020263671875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "5 Conclusion" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "" } ], [ { "bbox": { "b": 385.9627990722656, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 115.1929931640625, "t": 394.919189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "A" }, { "bbox": { "b": 385.9627990722656, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 358.81842041015625, "t": 394.919189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "Ethical Considerations and Broader Impacts Statement" }, { "bbox": { "b": 385.9627990722656, "coord_origin": "BOTTOMLEFT", "l": 494.040771484375, "r": 504.00335693359375, "t": 394.919189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "20" } ], [ { "bbox": { "b": 343.1530456542969, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 137.06100463867188, "t": 352.0596008300781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "B.1" }, { "bbox": { "b": 343.1530456542969, "coord_origin": "BOTTOMLEFT", "l": 145.8579864501953, "r": 479.1070556640625, "t": 352.0596008300781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "Pseudocode to recreate the text structure from the HTML code . . . . . . . . . . ." }, { "bbox": { "b": 343.1530456542969, "coord_origin": "BOTTOMLEFT", "l": 494.0409851074219, "r": 504.0035705566406, "t": 352.0596008300781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "21" } ], [ { "bbox": { "b": 326.7650451660156, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 137.06100463867188, "t": 335.6716003417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "B.2" }, { "bbox": { "b": 326.7650451660156, "coord_origin": "BOTTOMLEFT", "l": 145.8579864501953, "r": 479.0971374511719, "t": 335.6716003417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "Visualization tool use cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 326.7650451660156, "coord_origin": "BOTTOMLEFT", "l": 494.0410461425781, "r": 504.0036315917969, "t": 335.6716003417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "21" } ], [ { "bbox": { "b": 310.37603759765625, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 137.06100463867188, "t": 319.2825927734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "B.3" }, { "bbox": { "b": 310.37603759765625, "coord_origin": "BOTTOMLEFT", "l": 145.8579864501953, "r": 479.1070556640625, "t": 319.2825927734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "Exhaustive list of functions used in (Crowd)Sourced dataset . . . . . . . . . . . ." }, { "bbox": { "b": 310.37603759765625, "coord_origin": "BOTTOMLEFT", "l": 494.0409851074219, "r": 504.0035705566406, "t": 319.2825927734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "22" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "" }, { "bbox": { "b": 284.0947570800781, "coord_origin": "BOTTOMLEFT", "l": 494.04083251953125, "r": 504.00341796875, "t": 293.0511474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "26" } ], [ { "bbox": { "b": 284.0947570800781, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 115.1929931640625, "t": 293.0511474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "C" }, { "bbox": { "b": 284.0947570800781, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 362.41497802734375, "t": 293.0511474609375 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "Exhaustive list of human curated filters used on OSCAR" }, { "bbox": { "b": 284.0947570800781, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 362.41497802734375, "t": 293.0511474609375 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "Exhaustive list of human curated filters used on OSCAR" } ], [ { "bbox": { "b": 257.7437438964844, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 115.1929931640625, "t": 266.70013427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "D" }, { "bbox": { "b": 257.7437438964844, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 212.96595764160156, "t": 266.70013427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "PII filtering initiative" }, { "bbox": { "b": 257.7437438964844, "coord_origin": "BOTTOMLEFT", "l": 494.040771484375, "r": 504.00335693359375, "t": 266.70013427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "27" } ], [ { "bbox": { "b": 205.04074096679688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 212.7467803955078, "t": 240.34912109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "E F Author contributions" }, { "bbox": { "b": 231.3927459716797, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 178.9337158203125, "t": 240.34912109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "Data Sources" }, { "bbox": { "b": 205.04074096679688, "coord_origin": "BOTTOMLEFT", "l": 494.040771484375, "r": 504.00335693359375, "t": 240.34912109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "28 34" } ] ], "num_cols": 3, "num_rows": 23, "table_cells": [ { "bbox": { "b": 681.6038208007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 177.0109405517578, "t": 690.5601806640625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "1 Introduction" }, { "bbox": { "b": 681.6038208007812, "coord_origin": "BOTTOMLEFT", "l": 499.0220947265625, "r": 504.0033874511719, "t": 690.5601806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "3" }, { "bbox": { "b": 665.1460571289062, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 674.0526123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "1.1" }, { "bbox": { "b": 665.1460571289062, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970764160156, "t": 674.0526123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Outline of the Paper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 665.1460571289062, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 674.0526123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "3" }, { "bbox": { "b": 648.757080078125, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 657.6636352539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "1.2" }, { "bbox": { "b": 648.757080078125, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970458984375, "t": 657.6636352539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Related Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 648.757080078125, "coord_origin": "BOTTOMLEFT", "l": 499.0222473144531, "r": 504.0035400390625, "t": 657.6636352539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "4" }, { "bbox": { "b": 622.475830078125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 112.9813003540039, "t": 631.4321899414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "2" }, { "bbox": { "b": 622.475830078125, "coord_origin": "BOTTOMLEFT", "l": 122.94390106201172, "r": 504.0033874511719, "t": 631.4321899414062 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "(Crowd) Sourcing a Language Resource Catalogue 4" }, { "bbox": { "b": 606.01708984375, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 614.9236450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "2.1" }, { "bbox": { "b": 606.01708984375, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.1070251464844, "t": 614.9236450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "Obtaining Data from the Identified Resources . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 606.01708984375, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 614.9236450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "5" }, { "bbox": { "b": 589.6290893554688, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 598.53564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "2.2" }, { "bbox": { "b": 589.6290893554688, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.10699462890625, "t": 598.53564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Processing Pipeline for Quality Improvement on Crowdsourced Datasets . . . . . ." }, { "bbox": { "b": 589.6290893554688, "coord_origin": "BOTTOMLEFT", "l": 499.022216796875, "r": 504.0035095214844, "t": 598.53564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "6" }, { "bbox": { "b": 563.3478393554688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 572.30419921875 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "3 Processing OSCAR 7" }, { "bbox": { "b": 546.8890991210938, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 555.795654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "3.1" }, { "bbox": { "b": 546.8890991210938, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.09710693359375, "t": 555.795654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Data cleaning and filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 546.8890991210938, "coord_origin": "BOTTOMLEFT", "l": 499.0223083496094, "r": 504.00360107421875, "t": 555.795654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "7" }, { "bbox": { "b": 530.5010986328125, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 539.4076538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "3.2" }, { "bbox": { "b": 530.5010986328125, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970458984375, "t": 539.4076538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Deduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 530.5010986328125, "coord_origin": "BOTTOMLEFT", "l": 499.0222473144531, "r": 504.0035400390625, "t": 539.4076538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "8" }, { "bbox": { "b": 514.1121215820312, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 523.0186767578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "3.3" }, { "bbox": { "b": 514.1121215820312, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.1070556640625, "t": 523.0186767578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "Personally identifiable information . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 514.1121215820312, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 523.0186767578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "8" }, { "bbox": { "b": 487.8308410644531, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0033874511719, "t": 496.7872314453125 }, "col_span": 3, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "4 A First look at ROOTS 8" }, { "bbox": { "b": 471.3721008300781, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 480.2786560058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "4.1" }, { "bbox": { "b": 471.3721008300781, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0970764160156, "t": 480.2786560058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "Natural Languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 471.3721008300781, "coord_origin": "BOTTOMLEFT", "l": 499.02227783203125, "r": 504.0035705566406, "t": 480.2786560058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "8" }, { "bbox": { "b": 454.9841003417969, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 463.8906555175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "4.2" }, { "bbox": { "b": 454.9841003417969, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.0971374511719, "t": 463.8906555175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "Programming Languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 454.9841003417969, "coord_origin": "BOTTOMLEFT", "l": 499.0223388671875, "r": 504.0036315917969, "t": 463.8906555175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "9" }, { "bbox": { "b": 438.5950927734375, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 135.39724731445312, "t": 447.50164794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "4.3" }, { "bbox": { "b": 438.5950927734375, "coord_origin": "BOTTOMLEFT", "l": 145.85797119140625, "r": 479.1070251464844, "t": 447.50164794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "Tokenizer analysis of the component datasets . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 438.5950927734375, "coord_origin": "BOTTOMLEFT", "l": 494.04095458984375, "r": 504.0035400390625, "t": 447.50164794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "10" }, { "bbox": { "b": 412.3138122558594, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 170.55516052246094, "t": 421.27020263671875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "5 Conclusion" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 385.9627990722656, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 115.1929931640625, "t": 394.919189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "A" }, { "bbox": { "b": 385.9627990722656, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 358.81842041015625, "t": 394.919189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "Ethical Considerations and Broader Impacts Statement" }, { "bbox": { "b": 385.9627990722656, "coord_origin": "BOTTOMLEFT", "l": 494.040771484375, "r": 504.00335693359375, "t": 394.919189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "20" }, { "bbox": { "b": 343.1530456542969, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 137.06100463867188, "t": 352.0596008300781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "B.1" }, { "bbox": { "b": 343.1530456542969, "coord_origin": "BOTTOMLEFT", "l": 145.8579864501953, "r": 479.1070556640625, "t": 352.0596008300781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "Pseudocode to recreate the text structure from the HTML code . . . . . . . . . . ." }, { "bbox": { "b": 343.1530456542969, "coord_origin": "BOTTOMLEFT", "l": 494.0409851074219, "r": 504.0035705566406, "t": 352.0596008300781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "21" }, { "bbox": { "b": 326.7650451660156, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 137.06100463867188, "t": 335.6716003417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "B.2" }, { "bbox": { "b": 326.7650451660156, "coord_origin": "BOTTOMLEFT", "l": 145.8579864501953, "r": 479.0971374511719, "t": 335.6716003417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "Visualization tool use cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ." }, { "bbox": { "b": 326.7650451660156, "coord_origin": "BOTTOMLEFT", "l": 494.0410461425781, "r": 504.0036315917969, "t": 335.6716003417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "21" }, { "bbox": { "b": 310.37603759765625, "coord_origin": "BOTTOMLEFT", "l": 122.94400024414062, "r": 137.06100463867188, "t": 319.2825927734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "B.3" }, { "bbox": { "b": 310.37603759765625, "coord_origin": "BOTTOMLEFT", "l": 145.8579864501953, "r": 479.1070556640625, "t": 319.2825927734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "Exhaustive list of functions used in (Crowd)Sourced dataset . . . . . . . . . . . ." }, { "bbox": { "b": 310.37603759765625, "coord_origin": "BOTTOMLEFT", "l": 494.0409851074219, "r": 504.0035705566406, "t": 319.2825927734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "22" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "" }, { "bbox": { "b": 284.0947570800781, "coord_origin": "BOTTOMLEFT", "l": 494.04083251953125, "r": 504.00341796875, "t": 293.0511474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "26" }, { "bbox": { "b": 284.0947570800781, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 115.1929931640625, "t": 293.0511474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "C" }, { "bbox": { "b": 284.0947570800781, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 362.41497802734375, "t": 293.0511474609375 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "Exhaustive list of human curated filters used on OSCAR" }, { "bbox": { "b": 257.7437438964844, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 115.1929931640625, "t": 266.70013427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "D" }, { "bbox": { "b": 257.7437438964844, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 212.96595764160156, "t": 266.70013427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "PII filtering initiative" }, { "bbox": { "b": 257.7437438964844, "coord_origin": "BOTTOMLEFT", "l": 494.040771484375, "r": 504.00335693359375, "t": 266.70013427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "27" }, { "bbox": { "b": 205.04074096679688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 212.7467803955078, "t": 240.34912109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "E F Author contributions" }, { "bbox": { "b": 231.3927459716797, "coord_origin": "BOTTOMLEFT", "l": 122.94389343261719, "r": 178.9337158203125, "t": 240.34912109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "Data Sources" }, { "bbox": { "b": 205.04074096679688, "coord_origin": "BOTTOMLEFT", "l": 494.040771484375, "r": 504.00335693359375, "t": 240.34912109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "28 34" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 205.04074096679688, "coord_origin": "BOTTOMLEFT", "l": 107.0748519897461, "r": 504.34716796875, "t": 690.791015625 }, "charspan": [ 0, 0 ], "page_no": 2 } ], "references": [], "self_ref": "#/tables/0" }, { "captions": [ { "$ref": "#/texts/72" } ], "children": [], "data": { "grid": [ [ { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 124.81900024414062, "r": 138.65704345703125, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "AR" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 152.41539001464844, "r": 165.69554138183594, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "EU" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 179.4439239501953, "r": 193.281982421875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "BN" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 208.83360290527344, "r": 222.67166137695312, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "CA" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 238.50222778320312, "r": 251.78237915039062, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "ZH" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 267.89190673828125, "r": 281.17205810546875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "EN" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 297.82952880859375, "r": 310.0137939453125, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "FR" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 328.0560607910156, "r": 338.56658935546875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 0, "text": "HI" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 357.4457092285156, "r": 367.95623779296875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 0, "text": "ID" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 386.2774658203125, "r": 397.90380859375, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 0, "text": "PT" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 414.561279296875, "r": 428.3993225097656, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 0, "text": "UR" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 445.61468505859375, "r": 456.1252136230469, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 0, "text": "VI" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 474.4464416503906, "r": 486.0727844238281, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 0, "text": "ES" } ], [ { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 123.02100372314453, "r": 140.45555114746094, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "20.3" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 152.8291015625, "r": 165.2823486328125, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "5.2" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 177.64593505859375, "r": 195.0804901123047, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "48.8" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 207.03561401367188, "r": 224.4701690673828, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "21.1" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 236.42529296875, "r": 253.85984802246094, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "23.1" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 265.8149719238281, "r": 283.24951171875, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "17.2" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 295.2046203613281, "r": 312.63916015625, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "17.0" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 324.5942687988281, "r": 342.02880859375, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 1, "text": "25.7" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 353.9839172363281, "r": 371.41845703125, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 1, "text": "10.4" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 383.3735656738281, "r": 400.80810546875, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 1, "text": "12.6" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 412.7632141113281, "r": 430.19775390625, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 1, "text": "15.8" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 442.1528625488281, "r": 459.58740234375, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 1, "text": "21.3" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 471.5425109863281, "r": 488.97705078125, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 1, "text": "16.9" } ] ], "num_cols": 13, "num_rows": 2, "table_cells": [ { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 124.81900024414062, "r": 138.65704345703125, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "AR" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 152.41539001464844, "r": 165.69554138183594, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "EU" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 179.4439239501953, "r": 193.281982421875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "BN" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 208.83360290527344, "r": 222.67166137695312, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "CA" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 238.50222778320312, "r": 251.78237915039062, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "ZH" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 267.89190673828125, "r": 281.17205810546875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "EN" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 297.82952880859375, "r": 310.0137939453125, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "FR" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 328.0560607910156, "r": 338.56658935546875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 0, "text": "HI" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 357.4457092285156, "r": 367.95623779296875, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 0, "text": "ID" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 386.2774658203125, "r": 397.90380859375, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 0, "text": "PT" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 414.561279296875, "r": 428.3993225097656, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 0, "text": "UR" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 445.61468505859375, "r": 456.1252136230469, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 0, "text": "VI" }, { "bbox": { "b": 707.8211059570312, "coord_origin": "BOTTOMLEFT", "l": 474.4464416503906, "r": 486.0727844238281, "t": 716.7276611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 0, "text": "ES" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 123.02100372314453, "r": 140.45555114746094, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "20.3" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 152.8291015625, "r": 165.2823486328125, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "5.2" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 177.64593505859375, "r": 195.0804901123047, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "48.8" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 207.03561401367188, "r": 224.4701690673828, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "21.1" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 236.42529296875, "r": 253.85984802246094, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "23.1" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 265.8149719238281, "r": 283.24951171875, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "17.2" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 295.2046203613281, "r": 312.63916015625, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "17.0" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 324.5942687988281, "r": 342.02880859375, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 1, "text": "25.7" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 353.9839172363281, "r": 371.41845703125, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 1, "text": "10.4" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 383.3735656738281, "r": 400.80810546875, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 1, "text": "12.6" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 412.7632141113281, "r": 430.19775390625, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 1, "text": "15.8" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 442.1528625488281, "r": 459.58740234375, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 1, "text": "21.3" }, { "bbox": { "b": 692.1300659179688, "coord_origin": "BOTTOMLEFT", "l": 471.5425109863281, "r": 488.97705078125, "t": 701.03662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 1, "text": "16.9" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 689.6676025390625, "coord_origin": "BOTTOMLEFT", "l": 115.32803344726562, "r": 496.0636901855469, "t": 721.0944213867188 }, "charspan": [ 0, 0 ], "page_no": 8 } ], "references": [], "self_ref": "#/tables/1" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 686.0670776367188, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0020751953125, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Cettolo, M., C. Girardi, and M. Federico (2012, May 28-30). WIT3: Web inventory of transcribed and translated talks. In Proceedings of the 16th Annual conference of the European Association for Machine Translation , Trento, Italy, pp. 261-268. European Association for Machine Translation." } ], [ { "bbox": { "b": 645.3270874023438, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.24945068359375, "t": 676.0516357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Charikar, M. S. (2002). Similarity estimation techniques from rounding algorithms. In Proceedings of the Thiry-Fourth Annual ACM Symposium on Theory of Computing , STOC '02, New York, NY, USA, pp. 380-388. Association for Computing Machinery." } ], [ { "bbox": { "b": 604.5870971679688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74505615234375, "t": 635.3116455078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Chen, Y. and A. Eisele (2012, May). MultiUN v2: UN documents with multilingual alignments. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12) , Istanbul, Turkey, pp. 2500-2504. European Language Resources Association (ELRA)." } ], [ { "bbox": { "b": 574.756103515625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 503.9970703125, "t": 594.5716552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Clement, C. B., M. Bierbaum, K. P. O'Keeffe, and A. A. Alemi (2019). On the use of arxiv as a dataset. CoRR abs/1905.00075 ." } ], [ { "bbox": { "b": 523.1070556640625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.2471618652344, "t": 564.7406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Conneau, A., K. Khandelwal, N. Goyal, V. Chaudhary, G. Wenzek, F. Guzmán, E. Grave, M. Ott, L. Zettlemoyer, and V. Stoyanov (2020, July). Unsupervised cross-lingual representation learning at scale. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics , Online, pp. 8440-8451. Association for Computational Linguistics." } ], [ { "bbox": { "b": 504.1850891113281, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.00140380859375, "t": 513.0916748046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Contractor, D., D. McDuff, J. Haines, J. Lee, C. Hines, and B. Hecht (2020). Behavioral use licensing" } ], [ { "bbox": { "b": 474.35308837890625, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 373.6627197265625, "t": 483.2596435546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "David, D. (2020, December). Swahili: News classification dataset." } ], [ { "bbox": { "b": 444.5220947265625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74407958984375, "t": 464.337646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "De la Rosa, J., E. G. Ponferrada, M. Romero, P. Villegas, P. González de Prado Salas, and M. Grandury (2022). BERTIN: Efficient pre-training of a Spanish language model using perplexity sampling." } ], [ { "bbox": { "b": 403.7820739746094, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 503.9972229003906, "t": 423.5976257324219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Devlin, J., M. Chang, K. Lee, and K. Toutanova (2018). BERT: pre-training of deep bidirectional transformers for language understanding. CoRR abs/1810.04805 ." } ], [ { "bbox": { "b": 363.0420837402344, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74566650390625, "t": 393.7666320800781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Dodge, J., M. Sap, A. Marasovi'c, W. Agnew, G. Ilharco, D. Groeneveld, M. Mitchell, and M. Gardner (2021). Documenting large webtext corpora: A case study on the colossal clean crawled corpus. pp. 1286-1305." } ], [ { "bbox": { "b": 333.2110900878906, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.16998291015625, "t": 353.0266418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Einea, O., A. Elnagar, and R. Al Debsi (2019). Sanad: Single-label arabic news articles dataset for automatic text categorization. Data in Brief 25 , 104076." } ], [ { "bbox": { "b": 292.4710693359375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.7458801269531, "t": 323.1956481933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "El-Haj, M. (2020, May). Habibi - a multi dialect multi national Arabic song lyrics corpus. In Proceedings of the 12th Language Resources and Evaluation Conference , Marseille, France, pp. 1318-1326. European Language Resources Association." } ], [ { "bbox": { "b": 243.71807861328125, "coord_origin": "BOTTOMLEFT", "l": 330.2149963378906, "r": 464.888671875, "t": 252.6246337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "arXiv preprint arXiv:1611.04033 ." } ], [ { "bbox": { "b": 243.71807861328125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 326.6292419433594, "t": 252.6246337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "El-Khair, I. A. (2016). 1.5 billion words arabic corpus." } ], [ { "bbox": { "b": 173.14707946777344, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74359130859375, "t": 192.96263122558594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Fan, A., S. Bhosale, H. Schwenk, Z. Ma, A. El-Kishky, S. Goyal, M. Baines, O. Celebi, G. Wenzek, V. Chaudhary, et al. (2021). Beyond english-centric multilingual machine translation. J. Mach." } ], [ { "bbox": { "b": 162.23709106445312, "coord_origin": "BOTTOMLEFT", "l": 117.68399810791016, "r": 224.77113342285156, "t": 171.14364624023438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Learn. Res. 22 (107), 1-48." } ], [ { "bbox": { "b": 121.49707794189453, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.7449951171875, "t": 152.2216339111328 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Galliano, S., E. Geoffrois, G. Gravier, J.-F. Bonastre, D. Mostefa, and K. Choukri (2006, May). Corpus description of the ESTER evaluation campaign for the rich transcription of French broadcast news. In Proceedings of the Fifth International Conference on Language Resources and Evaluation" } ], [ { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.24713134765625, "t": 100.57264709472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Gao, L., S. Biderman, S. Black, L. Golding, T. Hoppe, C. Foster, J. Phang, H. He, A. Thite, N. Nabeshima, S. Presser, and C. Leahy (2020). The pile: An 800gb dataset of diverse text for language modeling. arXiv preprint arXiv:2101.00027 ." } ] ], "num_cols": 1, "num_rows": 18, "table_cells": [ { "bbox": { "b": 686.0670776367188, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0020751953125, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Cettolo, M., C. Girardi, and M. Federico (2012, May 28-30). WIT3: Web inventory of transcribed and translated talks. In Proceedings of the 16th Annual conference of the European Association for Machine Translation , Trento, Italy, pp. 261-268. European Association for Machine Translation." }, { "bbox": { "b": 645.3270874023438, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.24945068359375, "t": 676.0516357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Charikar, M. S. (2002). Similarity estimation techniques from rounding algorithms. In Proceedings of the Thiry-Fourth Annual ACM Symposium on Theory of Computing , STOC '02, New York, NY, USA, pp. 380-388. Association for Computing Machinery." }, { "bbox": { "b": 604.5870971679688, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74505615234375, "t": 635.3116455078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Chen, Y. and A. Eisele (2012, May). MultiUN v2: UN documents with multilingual alignments. In Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12) , Istanbul, Turkey, pp. 2500-2504. European Language Resources Association (ELRA)." }, { "bbox": { "b": 574.756103515625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 503.9970703125, "t": 594.5716552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Clement, C. B., M. Bierbaum, K. P. O'Keeffe, and A. A. Alemi (2019). On the use of arxiv as a dataset. CoRR abs/1905.00075 ." }, { "bbox": { "b": 523.1070556640625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.2471618652344, "t": 564.7406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Conneau, A., K. Khandelwal, N. Goyal, V. Chaudhary, G. Wenzek, F. Guzmán, E. Grave, M. Ott, L. Zettlemoyer, and V. Stoyanov (2020, July). Unsupervised cross-lingual representation learning at scale. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics , Online, pp. 8440-8451. Association for Computational Linguistics." }, { "bbox": { "b": 504.1850891113281, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.00140380859375, "t": 513.0916748046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Contractor, D., D. McDuff, J. Haines, J. Lee, C. Hines, and B. Hecht (2020). Behavioral use licensing" }, { "bbox": { "b": 474.35308837890625, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 373.6627197265625, "t": 483.2596435546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "David, D. (2020, December). Swahili: News classification dataset." }, { "bbox": { "b": 444.5220947265625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74407958984375, "t": 464.337646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "De la Rosa, J., E. G. Ponferrada, M. Romero, P. Villegas, P. González de Prado Salas, and M. Grandury (2022). BERTIN: Efficient pre-training of a Spanish language model using perplexity sampling." }, { "bbox": { "b": 403.7820739746094, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 503.9972229003906, "t": 423.5976257324219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Devlin, J., M. Chang, K. Lee, and K. Toutanova (2018). BERT: pre-training of deep bidirectional transformers for language understanding. CoRR abs/1810.04805 ." }, { "bbox": { "b": 363.0420837402344, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74566650390625, "t": 393.7666320800781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Dodge, J., M. Sap, A. Marasovi'c, W. Agnew, G. Ilharco, D. Groeneveld, M. Mitchell, and M. Gardner (2021). Documenting large webtext corpora: A case study on the colossal clean crawled corpus. pp. 1286-1305." }, { "bbox": { "b": 333.2110900878906, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.16998291015625, "t": 353.0266418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Einea, O., A. Elnagar, and R. Al Debsi (2019). Sanad: Single-label arabic news articles dataset for automatic text categorization. Data in Brief 25 , 104076." }, { "bbox": { "b": 292.4710693359375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.7458801269531, "t": 323.1956481933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "El-Haj, M. (2020, May). Habibi - a multi dialect multi national Arabic song lyrics corpus. In Proceedings of the 12th Language Resources and Evaluation Conference , Marseille, France, pp. 1318-1326. European Language Resources Association." }, { "bbox": { "b": 243.71807861328125, "coord_origin": "BOTTOMLEFT", "l": 330.2149963378906, "r": 464.888671875, "t": 252.6246337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "arXiv preprint arXiv:1611.04033 ." }, { "bbox": { "b": 243.71807861328125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 326.6292419433594, "t": 252.6246337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "El-Khair, I. A. (2016). 1.5 billion words arabic corpus." }, { "bbox": { "b": 173.14707946777344, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74359130859375, "t": 192.96263122558594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Fan, A., S. Bhosale, H. Schwenk, Z. Ma, A. El-Kishky, S. Goyal, M. Baines, O. Celebi, G. Wenzek, V. Chaudhary, et al. (2021). Beyond english-centric multilingual machine translation. J. Mach." }, { "bbox": { "b": 162.23709106445312, "coord_origin": "BOTTOMLEFT", "l": 117.68399810791016, "r": 224.77113342285156, "t": 171.14364624023438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Learn. Res. 22 (107), 1-48." }, { "bbox": { "b": 121.49707794189453, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.7449951171875, "t": 152.2216339111328 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Galliano, S., E. Geoffrois, G. Gravier, J.-F. Bonastre, D. Mostefa, and K. Choukri (2006, May). Corpus description of the ESTER evaluation campaign for the rich transcription of French broadcast news. In Proceedings of the Fifth International Conference on Language Resources and Evaluation" }, { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.24713134765625, "t": 100.57264709472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Gao, L., S. Biderman, S. Black, L. Golding, T. Hoppe, C. Foster, J. Phang, H. He, A. Thite, N. Nabeshima, S. Presser, and C. Leahy (2020). The pile: An 800gb dataset of diverse text for language modeling. arXiv preprint arXiv:2101.00027 ." } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.2330322265625, "coord_origin": "BOTTOMLEFT", "l": 106.54933166503906, "r": 506.707275390625, "t": 717.3917236328125 }, "charspan": [ 0, 0 ], "page_no": 14 } ], "references": [], "self_ref": "#/tables/2" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 707.8850708007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 450.9824523925781, "t": 716.7916259765625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Shikali, S. C. and M. Refuoe (2019, November). Language modeling data for swahili." }, { "bbox": { "b": 707.8850708007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 450.9824523925781, "t": 716.7916259765625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Shikali, S. C. and M. Refuoe (2019, November). Language modeling data for swahili." } ], [ { "bbox": { "b": 656.2791137695312, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.00335693359375, "t": 697.91259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Siripragada, S., J. Philip, V. P. Namboodiri, and C. V. Jawahar (2020, May). A multilingual parallel corpora collection effort for Indian languages. In Proceedings of the 12th Language Resources and Evaluation Conference , Marseille, France, pp. 3743-3751. European Language Resources Association." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "" } ], [ { "bbox": { "b": 626.4910888671875, "coord_origin": "BOTTOMLEFT", "l": 117.96299743652344, "r": 327.60467529296875, "t": 635.3976440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "machine learning. arXiv preprint arXiv:2007.02423 ." }, { "bbox": { "b": 637.4000854492188, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.1700744628906, "t": 646.306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Sloane, M., E. Moss, O. Awomolo, and L. Forlano (2020). Participation is not a design fix for" } ], [ { "bbox": { "b": 574.8840942382812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.2470703125, "t": 616.5176391601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Smith, S., M. Patwary, B. Norick, P. LeGresley, S. Rajbhandari, J. Casper, Z. Liu, S. Prabhumoye, G. Zerveas, V. Korthikanti, E. Zhang, R. Child, R. Y. Aminabadi, J. Bernauer, X. Song, M. Shoeybi, Y. He, M. Houston, S. Tiwary, and B. Catanzaro (2022). Using deepspeed and megatron to train megatron-turing nlg 530b, a large-scale generative language model. arXiv ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "" } ], [ { "bbox": { "b": 523.278076171875, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.24713134765625, "t": 564.91162109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Talat, Z., A. Névéol, S. Biderman, M. Clinciu, M. Dey, S. Longpre, S. Luccioni, M. Masoud, M. Mitchell, D. Radev, S. Sharma, A. Subramonian, J. Tae, S. Tan, D. Tunuguntla, and O. van der Wal (2022). You reap what you sow: On the challenges of bias evaluation under multilingual settings. In Challenges & Perspectives in Creating Large Language Models ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "" } ], [ { "bbox": { "b": 493.48907470703125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.6535949707031, "t": 513.3046264648438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Vuong, Q.-H., V.-P. La, T.-H. T. Nguyen, M.-H. Nguyen, T.-T. Le, and M.-T. Ho (2021). An ai- enabled approach in analyzing media data: An example from data on covid-19 news coverage in" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "" } ], [ { "bbox": { "b": 441.8830871582031, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.73846435546875, "t": 472.6076354980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Wenzek, G., M.-A. Lachaux, A. Conneau, V. Chaudhary, F. Guzmán, A. Joulin, and É. Grave (2020). Ccnet: Extracting high quality monolingual datasets from web crawl data. In Proceedings of The 12th Language Resources and Evaluation Conference , pp. 4003-4012." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "" } ], [ { "bbox": { "b": 423.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 398.8979187011719, "t": 431.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Wibowo, H. A. (2020). Recibrew. https://github.com/haryoa/ingredbrew." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "" } ], [ { "bbox": { "b": 404.1250915527344, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 475.0705261230469, "t": 413.0316467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Winner, L. (2017). Do artifacts have politics? In Computer Ethics , pp. 177-192. Routledge." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "" } ], [ { "bbox": { "b": 374.3360900878906, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74627685546875, "t": 394.1516418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Xue, L., N. Constant, A. Roberts, M. Kale, R. Al-Rfou, A. Siddhant, A. Barua, and C. Raffel (2020). mt5: A massively multilingual pre-trained text-to-text transformer. CoRR abs/2010.11934 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "" } ], [ { "bbox": { "b": 314.76007080078125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.1698913574219, "t": 323.6666259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Zerrouki, T. and A. Balla (2017). Tashkeela: Novel corpus of arabic vocalized texts, data for" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "" } ], [ { "bbox": { "b": 303.85107421875, "coord_origin": "BOTTOMLEFT", "l": 117.96299743652344, "r": 318.9268493652344, "t": 312.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "auto-diacritization systems. Data in brief 11 , 147." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "" } ], [ { "bbox": { "b": 274.06207275390625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0030517578125, "t": 293.87762451171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Zhang, B., P. Williams, I. Titov, and R. Sennrich (2020). Improving massively multilingual neural machine translation and zero-shot translation. arXiv preprint arXiv:2004.11867 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "" } ], [ { "bbox": { "b": 244.27407836914062, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 503.9972229003906, "t": 264.0896301269531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Zhang, C., D. Ippolito, K. Lee, M. Jagielski, F. Tramèr, and N. Carlini (2021). Counterfactual memorization in neural language models. arXiv preprint arXiv:2112.12938 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "" } ], [ { "bbox": { "b": 203.57708740234375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.2470397949219, "t": 234.3016357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Zhang, S., S. Roller, N. Goyal, M. Artetxe, M. Chen, S. Chen, C. Dewan, M. Diab, X. Li, X. V. Lin, et al. (2022). Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "" } ], [ { "bbox": { "b": 151.97007751464844, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0037536621094, "t": 193.6036376953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Ziemski, M., M. Junczys-Dowmunt, and B. Pouliquen (2016, May). The United Nations parallel corpus v1.0. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16) , Portorož, Slovenia, pp. 3530-3534. European Language Resources Association (ELRA)." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "" } ] ], "num_cols": 2, "num_rows": 16, "table_cells": [ { "bbox": { "b": 707.8850708007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 450.9824523925781, "t": 716.7916259765625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Shikali, S. C. and M. Refuoe (2019, November). Language modeling data for swahili." }, { "bbox": { "b": 656.2791137695312, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.00335693359375, "t": 697.91259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Siripragada, S., J. Philip, V. P. Namboodiri, and C. V. Jawahar (2020, May). A multilingual parallel corpora collection effort for Indian languages. In Proceedings of the 12th Language Resources and Evaluation Conference , Marseille, France, pp. 3743-3751. European Language Resources Association." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "" }, { "bbox": { "b": 626.4910888671875, "coord_origin": "BOTTOMLEFT", "l": 117.96299743652344, "r": 327.60467529296875, "t": 635.3976440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "machine learning. arXiv preprint arXiv:2007.02423 ." }, { "bbox": { "b": 637.4000854492188, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.1700744628906, "t": 646.306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Sloane, M., E. Moss, O. Awomolo, and L. Forlano (2020). Participation is not a design fix for" }, { "bbox": { "b": 574.8840942382812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.2470703125, "t": 616.5176391601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Smith, S., M. Patwary, B. Norick, P. LeGresley, S. Rajbhandari, J. Casper, Z. Liu, S. Prabhumoye, G. Zerveas, V. Korthikanti, E. Zhang, R. Child, R. Y. Aminabadi, J. Bernauer, X. Song, M. Shoeybi, Y. He, M. Houston, S. Tiwary, and B. Catanzaro (2022). Using deepspeed and megatron to train megatron-turing nlg 530b, a large-scale generative language model. arXiv ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 523.278076171875, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.24713134765625, "t": 564.91162109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Talat, Z., A. Névéol, S. Biderman, M. Clinciu, M. Dey, S. Longpre, S. Luccioni, M. Masoud, M. Mitchell, D. Radev, S. Sharma, A. Subramonian, J. Tae, S. Tan, D. Tunuguntla, and O. van der Wal (2022). You reap what you sow: On the challenges of bias evaluation under multilingual settings. In Challenges & Perspectives in Creating Large Language Models ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "" }, { "bbox": { "b": 493.48907470703125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.6535949707031, "t": 513.3046264648438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Vuong, Q.-H., V.-P. La, T.-H. T. Nguyen, M.-H. Nguyen, T.-T. Le, and M.-T. Ho (2021). An ai- enabled approach in analyzing media data: An example from data on covid-19 news coverage in" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "" }, { "bbox": { "b": 441.8830871582031, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.73846435546875, "t": 472.6076354980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Wenzek, G., M.-A. Lachaux, A. Conneau, V. Chaudhary, F. Guzmán, A. Joulin, and É. Grave (2020). Ccnet: Extracting high quality monolingual datasets from web crawl data. In Proceedings of The 12th Language Resources and Evaluation Conference , pp. 4003-4012." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "" }, { "bbox": { "b": 423.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 398.8979187011719, "t": 431.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Wibowo, H. A. (2020). Recibrew. https://github.com/haryoa/ingredbrew." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 404.1250915527344, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 475.0705261230469, "t": 413.0316467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Winner, L. (2017). Do artifacts have politics? In Computer Ethics , pp. 177-192. Routledge." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "" }, { "bbox": { "b": 374.3360900878906, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.74627685546875, "t": 394.1516418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Xue, L., N. Constant, A. Roberts, M. Kale, R. Al-Rfou, A. Siddhant, A. Barua, and C. Raffel (2020). mt5: A massively multilingual pre-trained text-to-text transformer. CoRR abs/2010.11934 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "" }, { "bbox": { "b": 314.76007080078125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.1698913574219, "t": 323.6666259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Zerrouki, T. and A. Balla (2017). Tashkeela: Novel corpus of arabic vocalized texts, data for" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "" }, { "bbox": { "b": 303.85107421875, "coord_origin": "BOTTOMLEFT", "l": 117.96299743652344, "r": 318.9268493652344, "t": 312.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "auto-diacritization systems. Data in brief 11 , 147." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 274.06207275390625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0030517578125, "t": 293.87762451171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Zhang, B., P. Williams, I. Titov, and R. Sennrich (2020). Improving massively multilingual neural machine translation and zero-shot translation. arXiv preprint arXiv:2004.11867 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "" }, { "bbox": { "b": 244.27407836914062, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 503.9972229003906, "t": 264.0896301269531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Zhang, C., D. Ippolito, K. Lee, M. Jagielski, F. Tramèr, and N. Carlini (2021). Counterfactual memorization in neural language models. arXiv preprint arXiv:2112.12938 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "" }, { "bbox": { "b": 203.57708740234375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 505.2470397949219, "t": 234.3016357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Zhang, S., S. Roller, N. Goyal, M. Artetxe, M. Chen, S. Chen, C. Dewan, M. Diab, X. Li, X. V. Lin, et al. (2022). Opt: Open pre-trained transformer language models. arXiv preprint arXiv:2205.01068 ." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 151.97007751464844, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 504.0037536621094, "t": 193.6036376953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Ziemski, M., M. Junczys-Dowmunt, and B. Pouliquen (2016, May). The United Nations parallel corpus v1.0. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16) , Portorož, Slovenia, pp. 3530-3534. European Language Resources Association (ELRA)." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 150.5994873046875, "coord_origin": "BOTTOMLEFT", "l": 106.56352233886719, "r": 506.5597839355469, "t": 716.7916259765625 }, "charspan": [ 0, 0 ], "page_no": 19 } ], "references": [], "self_ref": "#/tables/3" }, { "captions": [ { "$ref": "#/texts/258" } ], "children": [], "data": { "grid": [ [ { "bbox": { "b": 360.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Business Economy Technology Google services down for users around the world" }, { "bbox": { "b": 361.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 495.6666564941406, "t": 375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Business & Economy Technology Google services down for users around the world" } ], [ { "bbox": { "b": 327.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 136.6666717529297, "r": 304.3333435058594, "t": 359 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Frustrated customers countries including Australia, Japan , France and the United States complained online of the outage and tracking website DownDetector reported Google services were down in every continent" }, { "bbox": { "b": 327.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 330, "r": 497.6666564941406, "t": 359 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Frustrated customers countries including Australia, Japan , France and the United States complained online of the outage and tracking website DownDetector reported Google services were down in every continent." } ], [ { "bbox": { "b": 294.6666564941406, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 324.6666564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Popular Google services including Gmail and Drive were down for many users around the world Thursday , with the technology giant telling affected people they were aware of service disruption." }, { "bbox": { "b": 294, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 495.6666564941406, "t": 324.6666564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Popular Google services including Gmail and Drive were down for many users around the world Thursday , with the technology giant telling affected people they were aware of service disruption." } ], [ { "bbox": { "b": 277.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 293.6666564941406, "t": 292 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Anyone else having issues with @gmail in Australia?\" one person tweeted." }, { "bbox": { "b": 278, "coord_origin": "BOTTOMLEFT", "l": 332.6666564941406, "r": 487.6666564941406, "t": 292 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "Anyone else having issues with @gmail in Australia?\" one tweeted" } ], [ { "bbox": { "b": 252, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 307.6666564941406, "t": 275 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Another Twitter user, in Brooklyn, New York wrote: \"Nearly years and this the first time can remember Gmail being completely down" }, { "bbox": { "b": 252, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 501, "t": 282.6666564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "person Another Twitter user , in Brooklyn, New York wrote: \"Nearly years and this is the first time can remember Gmail being completely down ." } ], [ { "bbox": { "b": 228, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 299.6666564941406, "t": 250 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Google' s @Gmail Twitter feed replied to the posts with: Thanks for reporting. are aware service disruption the moment." }, { "bbox": { "b": 228, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 493, "t": 250 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Google @Gmail Twitter feed replied to the posts with: Thanks for reporting. are aware service disruption the moment" } ], [ { "bbox": { "b": 202.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 225.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "The message contained link Gmail service details page that told users are continuing investigate this issue and check back later" }, { "bbox": { "b": 202.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 330, "r": 495.6666564941406, "t": 225.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "The message contained link Gmail service details page that told users we are continuing investigate this issue and check back later_" } ], [ { "bbox": { "b": 185, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 199.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "well English, the Gail Twitter feed replied people in French, Japanese, Portuguese and German" }, { "bbox": { "b": 185.3333282470703, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 495.6666564941406, "t": 200 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "well English, the Gmail Twitter feed replied people in French, Japanese, Portuguese and German" } ], [ { "bbox": { "b": 168.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 294.3333435058594, "t": 183.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Responding AFP enquiry, Goog le refer the Suite Dashboard for status updates said" }, { "bbox": { "b": 168.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 487.6666564941406, "t": 183 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Responding AFP enquiry, Google said refer the Suite Dashboard for status updates" } ], [ { "bbox": { "b": 151.3333282470703, "coord_origin": "BOTTOMLEFT", "l": 136.3333282470703, "r": 263.6666564941406, "t": 166 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Talk show host Ellen DeGeneres apologizes over allegations" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "" } ], [ { "bbox": { "b": 135.3333282470703, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 288.3333435058594, "t": 148.6666717529297 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Trump VOwS to block any TikTok deal that allows Chinese control" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "" } ] ], "num_cols": 2, "num_rows": 11, "table_cells": [ { "bbox": { "b": 360.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Business Economy Technology Google services down for users around the world" }, { "bbox": { "b": 361.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 495.6666564941406, "t": 375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Business & Economy Technology Google services down for users around the world" }, { "bbox": { "b": 327.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 136.6666717529297, "r": 304.3333435058594, "t": 359 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Frustrated customers countries including Australia, Japan , France and the United States complained online of the outage and tracking website DownDetector reported Google services were down in every continent" }, { "bbox": { "b": 327.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 330, "r": 497.6666564941406, "t": 359 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Frustrated customers countries including Australia, Japan , France and the United States complained online of the outage and tracking website DownDetector reported Google services were down in every continent." }, { "bbox": { "b": 294.6666564941406, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 324.6666564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Popular Google services including Gmail and Drive were down for many users around the world Thursday , with the technology giant telling affected people they were aware of service disruption." }, { "bbox": { "b": 294, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 495.6666564941406, "t": 324.6666564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Popular Google services including Gmail and Drive were down for many users around the world Thursday , with the technology giant telling affected people they were aware of service disruption." }, { "bbox": { "b": 277.3333435058594, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 293.6666564941406, "t": 292 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Anyone else having issues with @gmail in Australia?\" one person tweeted." }, { "bbox": { "b": 278, "coord_origin": "BOTTOMLEFT", "l": 332.6666564941406, "r": 487.6666564941406, "t": 292 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "Anyone else having issues with @gmail in Australia?\" one tweeted" }, { "bbox": { "b": 252, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 307.6666564941406, "t": 275 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Another Twitter user, in Brooklyn, New York wrote: \"Nearly years and this the first time can remember Gmail being completely down" }, { "bbox": { "b": 252, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 501, "t": 282.6666564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "person Another Twitter user , in Brooklyn, New York wrote: \"Nearly years and this is the first time can remember Gmail being completely down ." }, { "bbox": { "b": 228, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 299.6666564941406, "t": 250 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Google' s @Gmail Twitter feed replied to the posts with: Thanks for reporting. are aware service disruption the moment." }, { "bbox": { "b": 228, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 493, "t": 250 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Google @Gmail Twitter feed replied to the posts with: Thanks for reporting. are aware service disruption the moment" }, { "bbox": { "b": 202.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 225.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "The message contained link Gmail service details page that told users are continuing investigate this issue and check back later" }, { "bbox": { "b": 202.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 330, "r": 495.6666564941406, "t": 225.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "The message contained link Gmail service details page that told users we are continuing investigate this issue and check back later_" }, { "bbox": { "b": 185, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 302.3333435058594, "t": 199.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "well English, the Gail Twitter feed replied people in French, Japanese, Portuguese and German" }, { "bbox": { "b": 185.3333282470703, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 495.6666564941406, "t": 200 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "well English, the Gmail Twitter feed replied people in French, Japanese, Portuguese and German" }, { "bbox": { "b": 168.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 294.3333435058594, "t": 183.3333282470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Responding AFP enquiry, Goog le refer the Suite Dashboard for status updates said" }, { "bbox": { "b": 168.6666717529297, "coord_origin": "BOTTOMLEFT", "l": 330.3333435058594, "r": 487.6666564941406, "t": 183 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Responding AFP enquiry, Google said refer the Suite Dashboard for status updates" }, { "bbox": { "b": 151.3333282470703, "coord_origin": "BOTTOMLEFT", "l": 136.3333282470703, "r": 263.6666564941406, "t": 166 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Talk show host Ellen DeGeneres apologizes over allegations" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "" }, { "bbox": { "b": 135.3333282470703, "coord_origin": "BOTTOMLEFT", "l": 137, "r": 288.3333435058594, "t": 148.6666717529297 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Trump VOwS to block any TikTok deal that allows Chinese control" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 116.98443603515625, "coord_origin": "BOTTOMLEFT", "l": 109.99005126953125, "r": 501.7123718261719, "t": 379.78912353515625 }, "charspan": [ 0, 0 ], "page_no": 23 } ], "references": [], "self_ref": "#/tables/4" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 185.99200439453125, "r": 201.68011474609375, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "ok_" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 214.0330047607422, "r": 229.7211151123047, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "_ok" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 242.07400512695312, "r": 257.7621154785156, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "k_o" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 270.114990234375, "r": 285.8031005859375, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "k_g" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 298.156005859375, "r": 313.8441162109375, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "_go" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 326.1969909667969, "r": 341.8851013183594, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "goo" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 354.2380065917969, "r": 369.9261169433594, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "ood" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 382.27899169921875, "r": 397.96710205078125, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 0, "text": "od_" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 410.32000732421875, "r": 426.00811767578125, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 0, "text": "d_o" } ], [ { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 191.34500122070312, "r": 196.3262939453125, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "2" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 219.38600158691406, "r": 224.36729431152344, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "2" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 247.427001953125, "r": 252.40829467773438, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 275.4679870605469, "r": 280.44927978515625, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 303.5090026855469, "r": 308.49029541015625, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 331.54998779296875, "r": 336.5312805175781, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 359.59100341796875, "r": 364.5722961425781, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 387.6319885253906, "r": 392.61328125, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 415.6730041503906, "r": 420.654296875, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 1, "text": "1" } ] ], "num_cols": 9, "num_rows": 2, "table_cells": [ { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 185.99200439453125, "r": 201.68011474609375, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "ok_" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 214.0330047607422, "r": 229.7211151123047, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "_ok" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 242.07400512695312, "r": 257.7621154785156, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "k_o" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 270.114990234375, "r": 285.8031005859375, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "k_g" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 298.156005859375, "r": 313.8441162109375, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "_go" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 326.1969909667969, "r": 341.8851013183594, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "goo" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 354.2380065917969, "r": 369.9261169433594, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "ood" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 382.27899169921875, "r": 397.96710205078125, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 0, "text": "od_" }, { "bbox": { "b": 135.1123046875, "coord_origin": "BOTTOMLEFT", "l": 410.32000732421875, "r": 426.00811767578125, "t": 143.41114807128906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 0, "text": "d_o" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 191.34500122070312, "r": 196.3262939453125, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "2" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 219.38600158691406, "r": 224.36729431152344, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "2" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 247.427001953125, "r": 252.40829467773438, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 275.4679870605469, "r": 280.44927978515625, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 303.5090026855469, "r": 308.49029541015625, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 331.54998779296875, "r": 336.5312805175781, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 359.59100341796875, "r": 364.5722961425781, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 387.6319885253906, "r": 392.61328125, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 1, "text": "1" }, { "bbox": { "b": 121.87207794189453, "coord_origin": "BOTTOMLEFT", "l": 415.6730041503906, "r": 420.654296875, "t": 130.7786407470703 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 1, "text": "1" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 119.85089111328125, "coord_origin": "BOTTOMLEFT", "l": 177.89434814453125, "r": 432.966796875, "t": 148.114013671875 }, "charspan": [ 0, 0 ], "page_no": 26 } ], "references": [], "self_ref": "#/tables/5" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 552.4798583984375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 153.51153564453125, "t": 561.4362182617188 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Dataset" }, { "bbox": { "b": 552.4798583984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 288.6710510253906, "t": 561.4362182617188 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 552.4798583984375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 400.9275817871094, "t": 561.4362182617188 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Source" } ], [ { "bbox": { "b": 534.1478271484375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 164.58993530273438, "t": 543.1041870117188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "AraBench" }, { "bbox": { "b": 534.0780639648438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 542.984619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "ar" }, { "bbox": { "b": 534.0780639648438, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 447.5924072265625, "t": 542.984619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "Sajjad et al. (2020)" } ], [ { "bbox": { "b": 509.9388122558594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 223.72792053222656, "t": 529.80419921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "1.5 billion words Arabic Corpus" }, { "bbox": { "b": 520.778076171875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 529.6846313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "ar" }, { "bbox": { "b": 520.778076171875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 435.68707275390625, "t": 529.6846313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "El-Khair (2016)" } ], [ { "bbox": { "b": 496.6388244628906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 167.3695068359375, "t": 505.59521484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "BanglaLM" }, { "bbox": { "b": 496.569091796875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.56158447265625, "t": 505.47564697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "bn" }, { "bbox": { "b": 496.569091796875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 458.0531311035156, "t": 505.47564697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "Kowsher et al. (2021)" } ], [ { "bbox": { "b": 472.4298095703125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 237.65565490722656, "t": 492.2951965332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "bangla sentiment classifica- tion datasets" }, { "bbox": { "b": 483.2690734863281, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.56158447265625, "t": 492.1756286621094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "bn" }, { "bbox": { "b": 483.2690734863281, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 455.8912658691406, "t": 492.1756286621094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "Rahman et al. (2018)" } ], [ { "bbox": { "b": 448.2208251953125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 239.05039978027344, "t": 468.0862121582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Question answering in Ben- gali" }, { "bbox": { "b": 459.0600891113281, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.56158447265625, "t": 467.9666442871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "bn" }, { "bbox": { "b": 459.0600891113281, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 463.07427978515625, "t": 467.9666442871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "Mayeesha et al. (2020)" } ], [ { "bbox": { "b": 434.9208068847656, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 211.5536346435547, "t": 443.877197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Binhvq News Corpus" }, { "bbox": { "b": 434.85107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 443.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "vi" }, { "bbox": { "b": 434.85107421875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 420.2052001953125, "t": 443.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "Bình (2021)" } ], [ { "bbox": { "b": 421.6208190917969, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 209.17257690429688, "t": 430.57720947265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Books by Book Dash" }, { "bbox": { "b": 421.55108642578125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 296.5515441894531, "t": 430.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "en, fr, xh, zu" }, { "bbox": { "b": 421.55108642578125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 482.5810546875, "t": 430.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "https://bookdash.org/books/" } ], [ { "bbox": { "b": 408.31982421875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 185.35198974609375, "t": 417.2762145996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Bloom Library" }, { "bbox": { "b": 397.3410949707031, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.63873291015625, "t": 417.1566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "ak, bm, fon, ki, lg, ln, nso, rw, st, sw, tn, ts, xh, zu" }, { "bbox": { "b": 408.2500915527344, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 438.7455749511719, "t": 417.1566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "bloomlibrary.org" } ], [ { "bbox": { "b": 384.1108093261719, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 164.58993530273438, "t": 393.06719970703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "BRAD 2.0" }, { "bbox": { "b": 384.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 392.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "ar" }, { "bbox": { "b": 384.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.6197814941406, "t": 392.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "Elnagar et al. (2018)" } ], [ { "bbox": { "b": 370.8108215332031, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 186.81649780273438, "t": 379.7672119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "brWaC Corpus" }, { "bbox": { "b": 370.7410888671875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 379.64764404296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "pt" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "" } ], [ { "bbox": { "b": 357.51080322265625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 164.0419921875, "t": 366.4671936035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "EusCrawl" }, { "bbox": { "b": 357.4410705566406, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 366.3476257324219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "eu" }, { "bbox": { "b": 357.4410705566406, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 452.414306640625, "t": 366.3476257324219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "Artetxe et al. (2022)" } ], [ { "bbox": { "b": 344.2108154296875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 234.32814025878906, "t": 353.1672058105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Catalan General Crawling" }, { "bbox": { "b": 344.1410827636719, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 353.0476379394531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "ca" }, { "bbox": { "b": 333.2320861816406, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 353.0476379394531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "Armengol-Estapé et al. (2021)" } ], [ { "bbox": { "b": 309.09283447265625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 211.00567626953125, "t": 328.9582214355469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Catalan Government Crawling" }, { "bbox": { "b": 319.9320983886719, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 328.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "ca" }, { "bbox": { "b": 309.0230712890625, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 328.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "Armengol-Estapé et al. (2021)" } ], [ { "bbox": { "b": 295.7928161621094, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 223.35928344726562, "t": 304.74920654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Catalan Textual Corpus" }, { "bbox": { "b": 295.72308349609375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 304.629638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "ca" }, { "bbox": { "b": 284.8130798339844, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 304.629638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "Armengol-Estapé et al. (2021)" } ], [ { "bbox": { "b": 260.673828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 228.55979919433594, "t": 280.5392150878906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Data on COVID-19 News Coverage in Vietnam" }, { "bbox": { "b": 271.5130920410156, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 280.4196472167969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "vi" }, { "bbox": { "b": 271.5130920410156, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 449.0569152832031, "t": 280.4196472167969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "Vuong et al. (2021)" } ], [ { "bbox": { "b": 247.37380981445312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 165.13787841796875, "t": 256.3302001953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "DuReader" }, { "bbox": { "b": 247.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 259.879150390625, "t": 256.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "zhs" }, { "bbox": { "b": 247.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 434.30230712890625, "t": 256.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "He et al. (2018)" } ], [ { "bbox": { "b": 223.16482543945312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 239.8673553466797, "t": 243.0301971435547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Enriched CONLLU Ancora for ML training" }, { "bbox": { "b": 234.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 242.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ca" }, { "bbox": { "b": 223.09507751464844, "coord_origin": "BOTTOMLEFT", "l": 371.4179992675781, "r": 473.5047607421875, "t": 242.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "Rodriguez-Penagos and Armentano-Oller (2021a)" } ], [ { "bbox": { "b": 209.8648223876953, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 154.08935546875, "t": 218.82119750976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "ESTER" }, { "bbox": { "b": 209.7950897216797, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 253.2340850830078, "t": 218.70164489746094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "fr" }, { "bbox": { "b": 209.7950897216797, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 456.9971008300781, "t": 218.70164489746094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "Galliano et al. (2006)" } ], [ { "bbox": { "b": 196.5648193359375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 151.87767028808594, "t": 205.5211944580078 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "Github" }, { "bbox": { "b": 196.49508666992188, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 265.40838623046875, "t": 205.40164184570312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "code" }, { "bbox": { "b": 196.49508666992188, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.68951416015625, "t": 205.40164184570312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "Github on BigQuery" } ], [ { "bbox": { "b": 183.2648162841797, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 150.77182006835938, "t": 192.22119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "Habibi" }, { "bbox": { "b": 183.19508361816406, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 192.1016387939453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "ar" }, { "bbox": { "b": 183.19508361816406, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.38824462890625, "t": 192.1016387939453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "El-Haj (2020)" } ], [ { "bbox": { "b": 169.96481323242188, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 143.01095581054688, "t": 178.9211883544922 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "HAL" }, { "bbox": { "b": 169.89508056640625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 253.2340850830078, "t": 178.8016357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "fr" }, { "bbox": { "b": 169.89508056640625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 468.0555725097656, "t": 178.8016357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "hal.archives-ouvertes.fr/" } ], [ { "bbox": { "b": 145.75482177734375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 236.03172302246094, "t": 165.6201934814453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "IIT Bombay English-Hindi Parallel Corpus" }, { "bbox": { "b": 156.59408569335938, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 165.50064086914062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "hi" }, { "bbox": { "b": 156.59408569335938, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 478.98455810546875, "t": 165.50064086914062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "Kunchukuttan et al. (2018)" } ], [ { "bbox": { "b": 132.45481872558594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 197.43661499023438, "t": 141.41119384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "IndicNLP Corpus" }, { "bbox": { "b": 121.47606658935547, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 362.0274353027344, "t": 141.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "bn, gu, hi, kn, ml, mr, or, pa, ta, te" }, { "bbox": { "b": 132.3850860595703, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 478.98455810546875, "t": 141.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "Kunchukuttan et al. (2020)" } ], [ { "bbox": { "b": 108.24581909179688, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 180.93856811523438, "t": 117.20218658447266 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "Indo4B BPPT" }, { "bbox": { "b": 108.17607879638672, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 117.0826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "id" }, { "bbox": { "b": 108.17607879638672, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 457.007080078125, "t": 117.0826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "Budiono et al. (2009)" } ], [ { "bbox": { "b": 94.94581604003906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 214.42286682128906, "t": 103.90219116210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "Indo4B OPUS JW300" }, { "bbox": { "b": 94.8760757446289, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 103.78263854980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "id" }, { "bbox": { "b": 94.8760757446289, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 460.9513244628906, "t": 103.78263854980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "Agi'c and Vuli'c (2019)" } ], [ { "bbox": { "b": 81.64581298828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 190.652099609375, "t": 90.60218811035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "Indo4B Kompas" }, { "bbox": { "b": 81.5760726928711, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 90.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "id" }, { "bbox": { "b": 81.5760726928711, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 443.16900634765625, "t": 90.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "Sakti et al. (2008)" } ], [ { "bbox": { "b": 68.34581756591797, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 222.52244567871094, "t": 77.30219268798828 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 27, "text": "Indo4B Parallel Corpus" }, { "bbox": { "b": 68.27607727050781, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 77.18264770507812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 27, "text": "id" }, { "bbox": { "b": 68.27607727050781, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 456.44915771484375, "t": 77.18264770507812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 27, "text": "Pisceldo et al. (2009)" } ] ], "num_cols": 3, "num_rows": 28, "table_cells": [ { "bbox": { "b": 552.4798583984375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 153.51153564453125, "t": 561.4362182617188 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Dataset" }, { "bbox": { "b": 552.4798583984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 288.6710510253906, "t": 561.4362182617188 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 552.4798583984375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 400.9275817871094, "t": 561.4362182617188 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Source" }, { "bbox": { "b": 534.1478271484375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 164.58993530273438, "t": 543.1041870117188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "AraBench" }, { "bbox": { "b": 534.0780639648438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 542.984619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "ar" }, { "bbox": { "b": 534.0780639648438, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 447.5924072265625, "t": 542.984619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "Sajjad et al. (2020)" }, { "bbox": { "b": 509.9388122558594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 223.72792053222656, "t": 529.80419921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "1.5 billion words Arabic Corpus" }, { "bbox": { "b": 520.778076171875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 529.6846313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "ar" }, { "bbox": { "b": 520.778076171875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 435.68707275390625, "t": 529.6846313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "El-Khair (2016)" }, { "bbox": { "b": 496.6388244628906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 167.3695068359375, "t": 505.59521484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "BanglaLM" }, { "bbox": { "b": 496.569091796875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.56158447265625, "t": 505.47564697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "bn" }, { "bbox": { "b": 496.569091796875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 458.0531311035156, "t": 505.47564697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "Kowsher et al. (2021)" }, { "bbox": { "b": 472.4298095703125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 237.65565490722656, "t": 492.2951965332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "bangla sentiment classifica- tion datasets" }, { "bbox": { "b": 483.2690734863281, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.56158447265625, "t": 492.1756286621094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "bn" }, { "bbox": { "b": 483.2690734863281, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 455.8912658691406, "t": 492.1756286621094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "Rahman et al. (2018)" }, { "bbox": { "b": 448.2208251953125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 239.05039978027344, "t": 468.0862121582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Question answering in Ben- gali" }, { "bbox": { "b": 459.0600891113281, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.56158447265625, "t": 467.9666442871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "bn" }, { "bbox": { "b": 459.0600891113281, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 463.07427978515625, "t": 467.9666442871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "Mayeesha et al. (2020)" }, { "bbox": { "b": 434.9208068847656, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 211.5536346435547, "t": 443.877197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Binhvq News Corpus" }, { "bbox": { "b": 434.85107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 443.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "vi" }, { "bbox": { "b": 434.85107421875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 420.2052001953125, "t": 443.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "Bình (2021)" }, { "bbox": { "b": 421.6208190917969, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 209.17257690429688, "t": 430.57720947265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Books by Book Dash" }, { "bbox": { "b": 421.55108642578125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 296.5515441894531, "t": 430.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "en, fr, xh, zu" }, { "bbox": { "b": 421.55108642578125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 482.5810546875, "t": 430.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "https://bookdash.org/books/" }, { "bbox": { "b": 408.31982421875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 185.35198974609375, "t": 417.2762145996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Bloom Library" }, { "bbox": { "b": 397.3410949707031, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.63873291015625, "t": 417.1566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "ak, bm, fon, ki, lg, ln, nso, rw, st, sw, tn, ts, xh, zu" }, { "bbox": { "b": 408.2500915527344, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 438.7455749511719, "t": 417.1566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "bloomlibrary.org" }, { "bbox": { "b": 384.1108093261719, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 164.58993530273438, "t": 393.06719970703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "BRAD 2.0" }, { "bbox": { "b": 384.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 392.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "ar" }, { "bbox": { "b": 384.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.6197814941406, "t": 392.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "Elnagar et al. (2018)" }, { "bbox": { "b": 370.8108215332031, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 186.81649780273438, "t": 379.7672119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "brWaC Corpus" }, { "bbox": { "b": 370.7410888671875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 379.64764404296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "pt" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "" }, { "bbox": { "b": 357.51080322265625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 164.0419921875, "t": 366.4671936035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "EusCrawl" }, { "bbox": { "b": 357.4410705566406, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 366.3476257324219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "eu" }, { "bbox": { "b": 357.4410705566406, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 452.414306640625, "t": 366.3476257324219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "Artetxe et al. (2022)" }, { "bbox": { "b": 344.2108154296875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 234.32814025878906, "t": 353.1672058105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Catalan General Crawling" }, { "bbox": { "b": 344.1410827636719, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 353.0476379394531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "ca" }, { "bbox": { "b": 333.2320861816406, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 353.0476379394531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "Armengol-Estapé et al. (2021)" }, { "bbox": { "b": 309.09283447265625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 211.00567626953125, "t": 328.9582214355469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Catalan Government Crawling" }, { "bbox": { "b": 319.9320983886719, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 328.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "ca" }, { "bbox": { "b": 309.0230712890625, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 328.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "Armengol-Estapé et al. (2021)" }, { "bbox": { "b": 295.7928161621094, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 223.35928344726562, "t": 304.74920654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Catalan Textual Corpus" }, { "bbox": { "b": 295.72308349609375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 304.629638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "ca" }, { "bbox": { "b": 284.8130798339844, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 304.629638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "Armengol-Estapé et al. (2021)" }, { "bbox": { "b": 260.673828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 228.55979919433594, "t": 280.5392150878906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Data on COVID-19 News Coverage in Vietnam" }, { "bbox": { "b": 271.5130920410156, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 280.4196472167969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "vi" }, { "bbox": { "b": 271.5130920410156, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 449.0569152832031, "t": 280.4196472167969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "Vuong et al. (2021)" }, { "bbox": { "b": 247.37380981445312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 165.13787841796875, "t": 256.3302001953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "DuReader" }, { "bbox": { "b": 247.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 259.879150390625, "t": 256.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "zhs" }, { "bbox": { "b": 247.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 434.30230712890625, "t": 256.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "He et al. (2018)" }, { "bbox": { "b": 223.16482543945312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 239.8673553466797, "t": 243.0301971435547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Enriched CONLLU Ancora for ML training" }, { "bbox": { "b": 234.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 242.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ca" }, { "bbox": { "b": 223.09507751464844, "coord_origin": "BOTTOMLEFT", "l": 371.4179992675781, "r": 473.5047607421875, "t": 242.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "Rodriguez-Penagos and Armentano-Oller (2021a)" }, { "bbox": { "b": 209.8648223876953, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 154.08935546875, "t": 218.82119750976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "ESTER" }, { "bbox": { "b": 209.7950897216797, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 253.2340850830078, "t": 218.70164489746094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "fr" }, { "bbox": { "b": 209.7950897216797, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 456.9971008300781, "t": 218.70164489746094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "Galliano et al. (2006)" }, { "bbox": { "b": 196.5648193359375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 151.87767028808594, "t": 205.5211944580078 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "Github" }, { "bbox": { "b": 196.49508666992188, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 265.40838623046875, "t": 205.40164184570312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "code" }, { "bbox": { "b": 196.49508666992188, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.68951416015625, "t": 205.40164184570312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "Github on BigQuery" }, { "bbox": { "b": 183.2648162841797, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 150.77182006835938, "t": 192.22119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "Habibi" }, { "bbox": { "b": 183.19508361816406, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 192.1016387939453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "ar" }, { "bbox": { "b": 183.19508361816406, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.38824462890625, "t": 192.1016387939453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "El-Haj (2020)" }, { "bbox": { "b": 169.96481323242188, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 143.01095581054688, "t": 178.9211883544922 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "HAL" }, { "bbox": { "b": 169.89508056640625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 253.2340850830078, "t": 178.8016357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "fr" }, { "bbox": { "b": 169.89508056640625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 468.0555725097656, "t": 178.8016357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "hal.archives-ouvertes.fr/" }, { "bbox": { "b": 145.75482177734375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 236.03172302246094, "t": 165.6201934814453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "IIT Bombay English-Hindi Parallel Corpus" }, { "bbox": { "b": 156.59408569335938, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 165.50064086914062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "hi" }, { "bbox": { "b": 156.59408569335938, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 478.98455810546875, "t": 165.50064086914062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "Kunchukuttan et al. (2018)" }, { "bbox": { "b": 132.45481872558594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 197.43661499023438, "t": 141.41119384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "IndicNLP Corpus" }, { "bbox": { "b": 121.47606658935547, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 362.0274353027344, "t": 141.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "bn, gu, hi, kn, ml, mr, or, pa, ta, te" }, { "bbox": { "b": 132.3850860595703, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 478.98455810546875, "t": 141.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "Kunchukuttan et al. (2020)" }, { "bbox": { "b": 108.24581909179688, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 180.93856811523438, "t": 117.20218658447266 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "Indo4B BPPT" }, { "bbox": { "b": 108.17607879638672, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 117.0826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "id" }, { "bbox": { "b": 108.17607879638672, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 457.007080078125, "t": 117.0826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "Budiono et al. (2009)" }, { "bbox": { "b": 94.94581604003906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 214.42286682128906, "t": 103.90219116210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "Indo4B OPUS JW300" }, { "bbox": { "b": 94.8760757446289, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 103.78263854980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "id" }, { "bbox": { "b": 94.8760757446289, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 460.9513244628906, "t": 103.78263854980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "Agi'c and Vuli'c (2019)" }, { "bbox": { "b": 81.64581298828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 190.652099609375, "t": 90.60218811035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "Indo4B Kompas" }, { "bbox": { "b": 81.5760726928711, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 90.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "id" }, { "bbox": { "b": 81.5760726928711, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 443.16900634765625, "t": 90.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "Sakti et al. (2008)" }, { "bbox": { "b": 68.34581756591797, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 222.52244567871094, "t": 77.30219268798828 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 27, "text": "Indo4B Parallel Corpus" }, { "bbox": { "b": 68.27607727050781, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 77.18264770507812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 27, "text": "id" }, { "bbox": { "b": 68.27607727050781, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 456.44915771484375, "t": 77.18264770507812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 27, "text": "Pisceldo et al. (2009)" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 65.74334716796875, "coord_origin": "BOTTOMLEFT", "l": 116.8939437866211, "r": 494.0702819824219, "t": 565.0388793945312 }, "charspan": [ 0, 0 ], "page_no": 28 } ], "references": [], "self_ref": "#/tables/6" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 153.51153564453125, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Dataset" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 288.6710510253906, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 400.9275817871094, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Source" } ], [ { "bbox": { "b": 689.9578247070312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 193.32208251953125, "t": 698.9141845703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Indo4B TALPCo" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "id" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 455.34332275390625, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "Nomoto et al. (2018)" } ], [ { "bbox": { "b": 676.6578369140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 184.44540405273438, "t": 685.6141967773438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Indo4B Tempo" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "id" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 443.16900634765625, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "Sakti et al. (2008)" } ], [ { "bbox": { "b": 652.4488525390625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 220.3207244873047, "t": 672.314208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Indonesian Frog Story- telling corpus" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "id" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 436.8028869628906, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "Moeljadi (2012)" } ], [ { "bbox": { "b": 628.2388916015625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 229.34683227539062, "t": 648.105224609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Indonesian News Articles Published at 2017" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "id" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.38824462890625, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "Ashari (2018)" } ], [ { "bbox": { "b": 614.9388427734375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 227.60337829589844, "t": 623.8952026367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Indonesian News Corpus" }, { "bbox": { "b": 614.8690795898438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 623.775634765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "id" }, { "bbox": { "b": 603.9600830078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 465.38800048828125, "t": 623.775634765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "Rahutomo and Miqdad Muadz Muzad (2018)" } ], [ { "bbox": { "b": 590.7298583984375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 159.07066345214844, "t": 599.6862182617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "IndoNLI" }, { "bbox": { "b": 590.6600952148438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 599.566650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "id" }, { "bbox": { "b": 590.6600952148438, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 463.07427978515625, "t": 599.566650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "Mahendra et al. (2021)" } ], [ { "bbox": { "b": 577.4298706054688, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 159.07066345214844, "t": 586.38623046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Indosum" }, { "bbox": { "b": 577.360107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 586.2666625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "id" }, { "bbox": { "b": 566.4511108398438, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 466.97320556640625, "t": 586.2666625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "Kurniawan and Louvan (2018)" } ], [ { "bbox": { "b": 553.2208251953125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 169.18270874023438, "t": 562.1771850585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "KALIMAT" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "ar" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 476.2946472167969, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "El-Haj and Koulali (2013)" } ], [ { "bbox": { "b": 539.9208374023438, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 163.48410034179688, "t": 548.877197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "KSUCCA" }, { "bbox": { "b": 539.85107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "ar" }, { "bbox": { "b": 539.85107421875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 457.5450439453125, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "Alrabiah et al. (2013)" } ], [ { "bbox": { "b": 526.620849609375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 149.0980987548828, "t": 535.5772094726562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "LABR" }, { "bbox": { "b": 526.5510864257812, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 535.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "ar" }, { "bbox": { "b": 526.5510864257812, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 457.27606201171875, "t": 535.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "Aly and Atiya (2013)" } ], [ { "bbox": { "b": 502.4118347167969, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 226.59715270996094, "t": 522.2772216796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Language modeling data for Swahili" }, { "bbox": { "b": 513.2510986328125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 257.6674499511719, "t": 522.1576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "sw" }, { "bbox": { "b": 513.2510986328125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 477.2012634277344, "t": 522.1576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "Shikali and Refuoe (2019)" } ], [ { "bbox": { "b": 489.1108093261719, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 237.54605102539062, "t": 498.06719970703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Leipzig Corpora Collection" }, { "bbox": { "b": 489.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.89784240722656, "t": 497.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "ur" }, { "bbox": { "b": 489.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 461.9783935546875, "t": 497.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "Goldhahn et al. (2012)" } ], [ { "bbox": { "b": 475.8108215332031, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 182.3134002685547, "t": 484.7672119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Mann Ki Baat" }, { "bbox": { "b": 464.83209228515625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 356.946533203125, "t": 484.64764404296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "bn, gu, hi, ml, mr, or, ta, te, ur" }, { "bbox": { "b": 475.7410888671875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 468.5636901855469, "t": 484.64764404296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "Siripragada et al. (2020)" } ], [ { "bbox": { "b": 451.601806640625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 175.11044311523438, "t": 460.5581970214844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Masakhaner" }, { "bbox": { "b": 451.5320739746094, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 329.32843017578125, "t": 460.4386291503906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "ig, lg, rw, sw, wo, yo" }, { "bbox": { "b": 451.5320739746094, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.2275085449219, "t": 460.4386291503906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "Adelani et al. (2021)" } ], [ { "bbox": { "b": 438.30181884765625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 172.0618896484375, "t": 447.2582092285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "MultiUN v2" }, { "bbox": { "b": 438.2320861816406, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 311.0870056152344, "t": 447.1386413574219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "en, ar, es, fr, zhs" }, { "bbox": { "b": 438.2320861816406, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 465.57489013671875, "t": 447.1386413574219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "Chen and Eisele (2012)" } ], [ { "bbox": { "b": 425.0018310546875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.71197509765625, "t": 433.9582214355469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Odiencorp" }, { "bbox": { "b": 424.9320983886719, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 269.28387451171875, "t": 433.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "en, or" }, { "bbox": { "b": 424.9320983886719, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 447.9909362792969, "t": 433.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "Parida et al. (2020)" } ], [ { "bbox": { "b": 411.7018127441406, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 200.02691650390625, "t": 420.658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Opensubtitles2016" }, { "bbox": { "b": 400.72308349609375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.89910888671875, "t": 420.53863525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ca, en, ar, es, eu, fr, id, bn, hi, ml, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 411.632080078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 487.36309814453125, "t": 420.53863525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "Lison and Tiedemann (2016)" } ], [ { "bbox": { "b": 387.4928283691406, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 180.74928283691406, "t": 396.44921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "OpenITI proc" }, { "bbox": { "b": 387.423095703125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 396.32965087890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "ar" }, { "bbox": { "b": 387.423095703125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 458.96966552734375, "t": 396.32965087890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "Belinkov et al. (2019)" } ], [ { "bbox": { "b": 374.19281005859375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.25369262695312, "t": 383.1492004394531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "OPUS-100" }, { "bbox": { "b": 352.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.60687255859375, "t": 383.0296325683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "ca, ar, eu, id, as, bn, gu, hi, ig, kn, ml, mr, or, pa, rw, ta, te, ur, pt, vi, xh, yo, zu" }, { "bbox": { "b": 374.1230773925781, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 448.1403503417969, "t": 383.0296325683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "Zhang et al. (2020)" } ], [ { "bbox": { "b": 339.0738220214844, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 189.2872314453125, "t": 348.03021240234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "ParlamentParla" }, { "bbox": { "b": 339.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 347.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "ca" }, { "bbox": { "b": 339.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.9461364746094, "t": 347.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "Külebi (2021)" } ], [ { "bbox": { "b": 325.7738037109375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 138.02964782714844, "t": 334.7301940917969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "PIB" }, { "bbox": { "b": 314.7950744628906, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 359.20245361328125, "t": 334.6106262207031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "bn, gu, hi, ml, mr, or, pa, ta, te, ur" }, { "bbox": { "b": 325.7040710449219, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 468.5636901855469, "t": 334.6106262207031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "Siripragada et al. (2020)" } ], [ { "bbox": { "b": 301.5648193359375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 200.5449676513672, "t": 310.5212097167969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "Project Gutenberg" }, { "bbox": { "b": 301.4950866699219, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 311.4954528808594, "t": 310.4016418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "en, es, fr, pt, zhs" }, { "bbox": { "b": 301.4950866699219, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.0295715332031, "t": 310.4016418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "gutenberg.org" } ], [ { "bbox": { "b": 277.3558044433594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 222.4327850341797, "t": 297.22119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "QED (formely AMARA Corpus)" }, { "bbox": { "b": 288.195068359375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 353.7070007324219, "t": 297.10162353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "ar, en, es, fr, hi, pt, zhs, zht" }, { "bbox": { "b": 288.195068359375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 456.9971008300781, "t": 297.10162353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "Abdelali et al. (2014)" } ], [ { "bbox": { "b": 264.0558166503906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 161.6310577392578, "t": 273.01220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "Recibrew" }, { "bbox": { "b": 263.986083984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 272.89263916015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "id" }, { "bbox": { "b": 263.986083984375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 434.3919372558594, "t": 272.89263916015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "Wibowo (2020)" } ], [ { "bbox": { "b": 250.7558135986328, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 232.29574584960938, "t": 259.7121887207031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "The Royal Society Corpus" }, { "bbox": { "b": 250.6860809326172, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 259.5926513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "en" }, { "bbox": { "b": 250.6860809326172, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.4205322265625, "t": 259.5926513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "Kermes et al. (2016)" } ], [ { "bbox": { "b": 237.45481872558594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 154.07940673828125, "t": 246.41119384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "S2ORC" }, { "bbox": { "b": 237.3850860595703, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 246.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "en" }, { "bbox": { "b": 237.3850860595703, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 433.7543640136719, "t": 246.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "Lo et al. (2020)" } ], [ { "bbox": { "b": 224.15481567382812, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 174.0045928955078, "t": 233.11119079589844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 27, "text": "Samanantar" }, { "bbox": { "b": 213.17608642578125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 360.8995056152344, "t": 232.99163818359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 27, "text": "as, bn, gu, hi, kn, ml, mr, or, pa, ta, te" }, { "bbox": { "b": 224.0850830078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.785400390625, "t": 232.99163818359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 27, "text": "Ramesh et al. (2021)" } ], [ { "bbox": { "b": 199.94581604003906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 155.533935546875, "t": 208.90219116210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 28, "text": "SANAD" }, { "bbox": { "b": 199.87608337402344, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 208.7826385498047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 28, "text": "ar" }, { "bbox": { "b": 199.87608337402344, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 445.3707580566406, "t": 208.7826385498047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 28, "text": "Einea et al. (2019)" } ], [ { "bbox": { "b": 186.64581298828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 155.19520568847656, "t": 195.60218811035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 29, "text": "SciELO" }, { "bbox": { "b": 186.57608032226562, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 282.0160827636719, "t": 195.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 29, "text": "en, es, pt" }, { "bbox": { "b": 186.57608032226562, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 410.6112365722656, "t": 195.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 29, "text": "scielo.org" } ], [ { "bbox": { "b": 173.34580993652344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 189.22744750976562, "t": 182.30218505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 30, "text": "Stack Exchange" }, { "bbox": { "b": 173.2760772705078, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 265.40838623046875, "t": 182.18263244628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 30, "text": "code" }, { "bbox": { "b": 173.2760772705078, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 439.2835998535156, "t": 182.18263244628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 30, "text": "Gao et al. (2020)" } ], [ { "bbox": { "b": 149.13682556152344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 237.66561889648438, "t": 169.002197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 31, "text": "Swahili News Classification Dataset" }, { "bbox": { "b": 159.97608947753906, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 257.6674499511719, "t": 168.8826446533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 31, "text": "sw" }, { "bbox": { "b": 159.97608947753906, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 424.977294921875, "t": 168.8826446533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 31, "text": "David (2020)" } ], [ { "bbox": { "b": 135.83682250976562, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 163.58370971679688, "t": 144.79319763183594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 32, "text": "Tashkeela" }, { "bbox": { "b": 135.76708984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 144.67364501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 32, "text": "ar" }, { "bbox": { "b": 135.76708984375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 476.0854187011719, "t": 144.67364501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 32, "text": "Zerrouki and Balla (2017)" } ], [ { "bbox": { "b": 122.53681945800781, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 150.3932342529297, "t": 131.49319458007812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 33, "text": "TeCLa" }, { "bbox": { "b": 122.46707916259766, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 131.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 33, "text": "ca" }, { "bbox": { "b": 122.46707916259766, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.12164306640625, "t": 131.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 33, "text": "Carrino et al. (2021)" } ], [ { "bbox": { "b": 109.2358169555664, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 145.87657165527344, "t": 119.77381896972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 34, "text": "WIT 3" }, { "bbox": { "b": 87.34807586669922, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.4580078125, "t": 118.07264709472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 34, "text": "ca, ar, en, es, eu, fr, id, as, bn, gu, hi, kn, ml, mr, pa, sw, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 109.16607666015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 452.0257568359375, "t": 118.07264709472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 34, "text": "Cettolo et al. (2012)" } ], [ { "bbox": { "b": 74.11781311035156, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 202.54745483398438, "t": 83.07418823242188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 35, "text": "The Pile: EuroParl" }, { "bbox": { "b": 74.0480728149414, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 293.2340087890625, "t": 82.95463562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 35, "text": "en, es, fr, pt" }, { "bbox": { "b": 74.0480728149414, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 439.2835998535156, "t": 82.95463562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 35, "text": "Gao et al. (2020)" } ] ], "num_cols": 3, "num_rows": 36, "table_cells": [ { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 153.51153564453125, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Dataset" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 288.6710510253906, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 400.9275817871094, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Source" }, { "bbox": { "b": 689.9578247070312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 193.32208251953125, "t": 698.9141845703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Indo4B TALPCo" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "id" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 455.34332275390625, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "Nomoto et al. (2018)" }, { "bbox": { "b": 676.6578369140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 184.44540405273438, "t": 685.6141967773438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Indo4B Tempo" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "id" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 443.16900634765625, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "Sakti et al. (2008)" }, { "bbox": { "b": 652.4488525390625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 220.3207244873047, "t": 672.314208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Indonesian Frog Story- telling corpus" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "id" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 436.8028869628906, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "Moeljadi (2012)" }, { "bbox": { "b": 628.2388916015625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 229.34683227539062, "t": 648.105224609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Indonesian News Articles Published at 2017" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "id" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.38824462890625, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "Ashari (2018)" }, { "bbox": { "b": 614.9388427734375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 227.60337829589844, "t": 623.8952026367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Indonesian News Corpus" }, { "bbox": { "b": 614.8690795898438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 623.775634765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "id" }, { "bbox": { "b": 603.9600830078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 465.38800048828125, "t": 623.775634765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "Rahutomo and Miqdad Muadz Muzad (2018)" }, { "bbox": { "b": 590.7298583984375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 159.07066345214844, "t": 599.6862182617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "IndoNLI" }, { "bbox": { "b": 590.6600952148438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 599.566650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "id" }, { "bbox": { "b": 590.6600952148438, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 463.07427978515625, "t": 599.566650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "Mahendra et al. (2021)" }, { "bbox": { "b": 577.4298706054688, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 159.07066345214844, "t": 586.38623046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Indosum" }, { "bbox": { "b": 577.360107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 586.2666625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "id" }, { "bbox": { "b": 566.4511108398438, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 466.97320556640625, "t": 586.2666625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "Kurniawan and Louvan (2018)" }, { "bbox": { "b": 553.2208251953125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 169.18270874023438, "t": 562.1771850585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "KALIMAT" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "ar" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 476.2946472167969, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "El-Haj and Koulali (2013)" }, { "bbox": { "b": 539.9208374023438, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 163.48410034179688, "t": 548.877197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "KSUCCA" }, { "bbox": { "b": 539.85107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "ar" }, { "bbox": { "b": 539.85107421875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 457.5450439453125, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "Alrabiah et al. (2013)" }, { "bbox": { "b": 526.620849609375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 149.0980987548828, "t": 535.5772094726562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "LABR" }, { "bbox": { "b": 526.5510864257812, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 535.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "ar" }, { "bbox": { "b": 526.5510864257812, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 457.27606201171875, "t": 535.4576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "Aly and Atiya (2013)" }, { "bbox": { "b": 502.4118347167969, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 226.59715270996094, "t": 522.2772216796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Language modeling data for Swahili" }, { "bbox": { "b": 513.2510986328125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 257.6674499511719, "t": 522.1576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "sw" }, { "bbox": { "b": 513.2510986328125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 477.2012634277344, "t": 522.1576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "Shikali and Refuoe (2019)" }, { "bbox": { "b": 489.1108093261719, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 237.54605102539062, "t": 498.06719970703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Leipzig Corpora Collection" }, { "bbox": { "b": 489.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.89784240722656, "t": 497.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "ur" }, { "bbox": { "b": 489.04107666015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 461.9783935546875, "t": 497.9476318359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "Goldhahn et al. (2012)" }, { "bbox": { "b": 475.8108215332031, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 182.3134002685547, "t": 484.7672119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Mann Ki Baat" }, { "bbox": { "b": 464.83209228515625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 356.946533203125, "t": 484.64764404296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "bn, gu, hi, ml, mr, or, ta, te, ur" }, { "bbox": { "b": 475.7410888671875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 468.5636901855469, "t": 484.64764404296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "Siripragada et al. (2020)" }, { "bbox": { "b": 451.601806640625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 175.11044311523438, "t": 460.5581970214844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Masakhaner" }, { "bbox": { "b": 451.5320739746094, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 329.32843017578125, "t": 460.4386291503906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "ig, lg, rw, sw, wo, yo" }, { "bbox": { "b": 451.5320739746094, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.2275085449219, "t": 460.4386291503906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "Adelani et al. (2021)" }, { "bbox": { "b": 438.30181884765625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 172.0618896484375, "t": 447.2582092285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "MultiUN v2" }, { "bbox": { "b": 438.2320861816406, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 311.0870056152344, "t": 447.1386413574219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "en, ar, es, fr, zhs" }, { "bbox": { "b": 438.2320861816406, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 465.57489013671875, "t": 447.1386413574219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "Chen and Eisele (2012)" }, { "bbox": { "b": 425.0018310546875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.71197509765625, "t": 433.9582214355469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Odiencorp" }, { "bbox": { "b": 424.9320983886719, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 269.28387451171875, "t": 433.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "en, or" }, { "bbox": { "b": 424.9320983886719, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 447.9909362792969, "t": 433.8386535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "Parida et al. (2020)" }, { "bbox": { "b": 411.7018127441406, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 200.02691650390625, "t": 420.658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Opensubtitles2016" }, { "bbox": { "b": 400.72308349609375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.89910888671875, "t": 420.53863525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ca, en, ar, es, eu, fr, id, bn, hi, ml, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 411.632080078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 487.36309814453125, "t": 420.53863525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "Lison and Tiedemann (2016)" }, { "bbox": { "b": 387.4928283691406, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 180.74928283691406, "t": 396.44921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "OpenITI proc" }, { "bbox": { "b": 387.423095703125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 396.32965087890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "ar" }, { "bbox": { "b": 387.423095703125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 458.96966552734375, "t": 396.32965087890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "Belinkov et al. (2019)" }, { "bbox": { "b": 374.19281005859375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.25369262695312, "t": 383.1492004394531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "OPUS-100" }, { "bbox": { "b": 352.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.60687255859375, "t": 383.0296325683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "ca, ar, eu, id, as, bn, gu, hi, ig, kn, ml, mr, or, pa, rw, ta, te, ur, pt, vi, xh, yo, zu" }, { "bbox": { "b": 374.1230773925781, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 448.1403503417969, "t": 383.0296325683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "Zhang et al. (2020)" }, { "bbox": { "b": 339.0738220214844, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 189.2872314453125, "t": 348.03021240234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "ParlamentParla" }, { "bbox": { "b": 339.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 347.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "ca" }, { "bbox": { "b": 339.00408935546875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.9461364746094, "t": 347.91064453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "Külebi (2021)" }, { "bbox": { "b": 325.7738037109375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 138.02964782714844, "t": 334.7301940917969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "PIB" }, { "bbox": { "b": 314.7950744628906, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 359.20245361328125, "t": 334.6106262207031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "bn, gu, hi, ml, mr, or, pa, ta, te, ur" }, { "bbox": { "b": 325.7040710449219, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 468.5636901855469, "t": 334.6106262207031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "Siripragada et al. (2020)" }, { "bbox": { "b": 301.5648193359375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 200.5449676513672, "t": 310.5212097167969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "Project Gutenberg" }, { "bbox": { "b": 301.4950866699219, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 311.4954528808594, "t": 310.4016418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "en, es, fr, pt, zhs" }, { "bbox": { "b": 301.4950866699219, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.0295715332031, "t": 310.4016418457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "gutenberg.org" }, { "bbox": { "b": 277.3558044433594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 222.4327850341797, "t": 297.22119140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "QED (formely AMARA Corpus)" }, { "bbox": { "b": 288.195068359375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 353.7070007324219, "t": 297.10162353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "ar, en, es, fr, hi, pt, zhs, zht" }, { "bbox": { "b": 288.195068359375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 456.9971008300781, "t": 297.10162353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "Abdelali et al. (2014)" }, { "bbox": { "b": 264.0558166503906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 161.6310577392578, "t": 273.01220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "Recibrew" }, { "bbox": { "b": 263.986083984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 272.89263916015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "id" }, { "bbox": { "b": 263.986083984375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 434.3919372558594, "t": 272.89263916015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "Wibowo (2020)" }, { "bbox": { "b": 250.7558135986328, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 232.29574584960938, "t": 259.7121887207031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "The Royal Society Corpus" }, { "bbox": { "b": 250.6860809326172, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 259.5926513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "en" }, { "bbox": { "b": 250.6860809326172, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.4205322265625, "t": 259.5926513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "Kermes et al. (2016)" }, { "bbox": { "b": 237.45481872558594, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 154.07940673828125, "t": 246.41119384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "S2ORC" }, { "bbox": { "b": 237.3850860595703, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 246.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "en" }, { "bbox": { "b": 237.3850860595703, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 433.7543640136719, "t": 246.29164123535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "Lo et al. (2020)" }, { "bbox": { "b": 224.15481567382812, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 174.0045928955078, "t": 233.11119079589844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 27, "text": "Samanantar" }, { "bbox": { "b": 213.17608642578125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 360.8995056152344, "t": 232.99163818359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 27, "text": "as, bn, gu, hi, kn, ml, mr, or, pa, ta, te" }, { "bbox": { "b": 224.0850830078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.785400390625, "t": 232.99163818359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 27, "text": "Ramesh et al. (2021)" }, { "bbox": { "b": 199.94581604003906, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 155.533935546875, "t": 208.90219116210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 28, "text": "SANAD" }, { "bbox": { "b": 199.87608337402344, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 208.7826385498047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 28, "text": "ar" }, { "bbox": { "b": 199.87608337402344, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 445.3707580566406, "t": 208.7826385498047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 28, "text": "Einea et al. (2019)" }, { "bbox": { "b": 186.64581298828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 155.19520568847656, "t": 195.60218811035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 29, "text": "SciELO" }, { "bbox": { "b": 186.57608032226562, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 282.0160827636719, "t": 195.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 29, "text": "en, es, pt" }, { "bbox": { "b": 186.57608032226562, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 410.6112365722656, "t": 195.48263549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 29, "text": "scielo.org" }, { "bbox": { "b": 173.34580993652344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 189.22744750976562, "t": 182.30218505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 30, "text": "Stack Exchange" }, { "bbox": { "b": 173.2760772705078, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 265.40838623046875, "t": 182.18263244628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 30, "text": "code" }, { "bbox": { "b": 173.2760772705078, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 439.2835998535156, "t": 182.18263244628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 30, "text": "Gao et al. (2020)" }, { "bbox": { "b": 149.13682556152344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 237.66561889648438, "t": 169.002197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 31, "text": "Swahili News Classification Dataset" }, { "bbox": { "b": 159.97608947753906, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 257.6674499511719, "t": 168.8826446533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 31, "text": "sw" }, { "bbox": { "b": 159.97608947753906, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 424.977294921875, "t": 168.8826446533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 31, "text": "David (2020)" }, { "bbox": { "b": 135.83682250976562, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 163.58370971679688, "t": 144.79319763183594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 32, "text": "Tashkeela" }, { "bbox": { "b": 135.76708984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.33993530273438, "t": 144.67364501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 32, "text": "ar" }, { "bbox": { "b": 135.76708984375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 476.0854187011719, "t": 144.67364501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 32, "text": "Zerrouki and Balla (2017)" }, { "bbox": { "b": 122.53681945800781, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 150.3932342529297, "t": 131.49319458007812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 33, "text": "TeCLa" }, { "bbox": { "b": 122.46707916259766, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 131.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 33, "text": "ca" }, { "bbox": { "b": 122.46707916259766, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 453.12164306640625, "t": 131.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 33, "text": "Carrino et al. (2021)" }, { "bbox": { "b": 109.2358169555664, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 145.87657165527344, "t": 119.77381896972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 34, "text": "WIT 3" }, { "bbox": { "b": 87.34807586669922, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.4580078125, "t": 118.07264709472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 34, "text": "ca, ar, en, es, eu, fr, id, as, bn, gu, hi, kn, ml, mr, pa, sw, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 109.16607666015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 452.0257568359375, "t": 118.07264709472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 34, "text": "Cettolo et al. (2012)" }, { "bbox": { "b": 74.11781311035156, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 202.54745483398438, "t": 83.07418823242188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 35, "text": "The Pile: EuroParl" }, { "bbox": { "b": 74.0480728149414, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 293.2340087890625, "t": 82.95463562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 35, "text": "en, es, fr, pt" }, { "bbox": { "b": 74.0480728149414, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 439.2835998535156, "t": 82.95463562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 35, "text": "Gao et al. (2020)" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 71.96478271484375, "coord_origin": "BOTTOMLEFT", "l": 117.16191101074219, "r": 494.06329345703125, "t": 720.1602172851562 }, "charspan": [ 0, 0 ], "page_no": 29 } ], "references": [], "self_ref": "#/tables/7" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 153.51153564453125, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Dataset" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 288.6710510253906, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 400.9275817871094, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Source" } ], [ { "bbox": { "b": 689.9578247070312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 196.01197814941406, "t": 698.9141845703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "The Pile: USPTO" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "en" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 439.2835998535156, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "Gao et al. (2020)" } ], [ { "bbox": { "b": 676.6578369140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 177.5114288330078, "t": 685.6141967773438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "UIT-VSMEC" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "vi" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 434.8601989746094, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "Ho et al. (2020)" } ], [ { "bbox": { "b": 652.4488525390625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 220.94834899902344, "t": 672.314208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "United Nations Parallel Corpus" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 311.0870056152344, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "ar, en, es, fr, zhs" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 455.34332275390625, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "Ziemski et al. (2016)" } ], [ { "bbox": { "b": 606.4208984375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 238.77146911621094, "t": 648.105224609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Unsupervised Cross-lingual Representation Learning at Scale Common Crawl Corpus" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "ne" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 458.10296630859375, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "Conneau et al. (2020)" } ], [ { "bbox": { "b": 593.120849609375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 234.8063201904297, "t": 602.0772094726562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Urdu Monolingual Corpus" }, { "bbox": { "b": 593.0510864257812, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.89784240722656, "t": 601.9576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "ur" }, { "bbox": { "b": 593.0510864257812, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 450.10296630859375, "t": 601.9576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "Jawaid et al. (2014)" } ], [ { "bbox": { "b": 579.8208618164062, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 170.746826171875, "t": 588.7772216796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "VietAI SAT" }, { "bbox": { "b": 579.7510986328125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 588.6576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "vi" }, { "bbox": { "b": 579.7510986328125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 459.1390686035156, "t": 588.6576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "Ngo and Trinh (2021)" } ], [ { "bbox": { "b": 566.5208740234375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 234.74655151367188, "t": 575.4772338867188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Vietnamese Poetry Corpus" }, { "bbox": { "b": 566.4511108398438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 575.357666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "vi" }, { "bbox": { "b": 566.4511108398438, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.2275085449219, "t": 575.357666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "Nguyen et al. (2021)" } ], [ { "bbox": { "b": 553.2208251953125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 167.548828125, "t": 562.1771850585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "UIT-VSFC" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "vi" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.2275085449219, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "Nguyen et al. (2018)" } ], [ { "bbox": { "b": 539.9208374023438, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.44297790527344, "t": 548.877197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "VilaQuAD" }, { "bbox": { "b": 539.85107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "ca" }, { "bbox": { "b": 528.9420776367188, "coord_origin": "BOTTOMLEFT", "l": 371.4179992675781, "r": 474.0626525878906, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "Rodriguez-Penagos and Armentano-Oller (2021b)" } ], [ { "bbox": { "b": 504.8028259277344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 220.96827697753906, "t": 524.668212890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "VinBigdata-VSLP ASR Challenge 2020" }, { "bbox": { "b": 515.64208984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 524.5486450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "vi" }, { "bbox": { "b": 493.8230895996094, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 536.6311645507812, "t": 524.5486450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "institute.vinbigdata.org/ev ents/vinbigdata- chia-se-100-gio-du-lieu- tieng-noi-cho-cong-dong/" } ], [ { "bbox": { "b": 469.6838073730469, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 240.3455352783203, "t": 489.5491943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "VinBigdata-VSLP Monolin- gual Corpus 2020" }, { "bbox": { "b": 480.5230712890625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 489.42962646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "vi" }, { "bbox": { "b": 458.705078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 536.6311645507812, "t": 489.42962646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "institute.vinbigdata.org/ev ents/vinbigdata- chia-se-100-gio-du-lieu- tieng-noi-cho-cong-dong/" } ], [ { "bbox": { "b": 434.5658264160156, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 239.80755615234375, "t": 454.43121337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "VinBigdata-VSLP Bilingual Corpus 2020" }, { "bbox": { "b": 445.40509033203125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 454.3116455078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "vi" }, { "bbox": { "b": 423.58709716796875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 536.6311645507812, "t": 454.3116455078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "institute.vinbigdata.org/ev ents/vinbigdata- chia-se-100-gio-du-lieu- tieng-noi-cho-cong-dong/" } ], [ { "bbox": { "b": 410.3568115234375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 172.5400848388672, "t": 419.3132019042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "ViquiQuAD" }, { "bbox": { "b": 410.2870788574219, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 419.1936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "ca" }, { "bbox": { "b": 399.3780822753906, "coord_origin": "BOTTOMLEFT", "l": 371.4179992675781, "r": 473.5047607421875, "t": 419.1936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "Rodriguez-Penagos and Armentano-Oller (2021c)" } ], [ { "bbox": { "b": 386.1478271484375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 198.8114776611328, "t": 395.1042175292969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "VNTQcorpus(big)" }, { "bbox": { "b": 386.0780944824219, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 394.9846496582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "vi" }, { "bbox": { "b": 375.1680908203125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 491.60552978515625, "t": 394.9846496582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "http://viet.jnlp.org/download- du-lieu-tu-vung-corpus" } ], [ { "bbox": { "b": 361.93780517578125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 167.2001495361328, "t": 370.8941955566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Wikibooks" }, { "bbox": { "b": 340.0500793457031, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.89910888671875, "t": 370.7746276855469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "ca, ar, en, es, eu, fr, id, bn, hi, ml, mr, pa, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 361.8680725097656, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 428.88262939453125, "t": 370.7746276855469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "wikibooks.org" } ], [ { "bbox": { "b": 326.81982421875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 168.29603576660156, "t": 335.7762145996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Wikimedia" }, { "bbox": { "b": 326.7500915527344, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 293.6424255371094, "t": 335.6566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "ca, id, hi, pt" }, { "bbox": { "b": 326.7500915527344, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 429.4305725097656, "t": 335.6566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "wikimedia.org" } ], [ { "bbox": { "b": 313.5198059082031, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 163.31472778320312, "t": 322.4761962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Wikinews" }, { "bbox": { "b": 313.4500732421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 349.8216247558594, "t": 322.35662841796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ar, ca, en, es, fr, ta, pt, zhs" }, { "bbox": { "b": 313.4500732421875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 425.3060607910156, "t": 322.35662841796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "wikinews.org" } ], [ { "bbox": { "b": 300.2198181152344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 165.53639221191406, "t": 309.17620849609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "Wikipedia" }, { "bbox": { "b": 256.5140686035156, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.2080993652344, "t": 309.056640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "ak, ar, as, bm, bn, ca, en, es, eu, fr, id, ig, gu, hi, ki, kn, lg, ln, ml, mr, nso, ny, or, pa, pt, rn, rw, sn, st, sw, ta, te, tn, ts, tum, tw, ur, vi, wo, yo, zhs," }, { "bbox": { "b": 300.15008544921875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 426.6609802246094, "t": 309.056640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "wikipedia.org" } ], [ { "bbox": { "b": 232.37380981445312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.08433532714844, "t": 241.33018493652344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "Wikiquote" }, { "bbox": { "b": 210.486083984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.8990783691406, "t": 241.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "ar, ca, en, es, eu, fr, id, gu, hi, kn, ml, mr, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 232.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.2188720703125, "t": 241.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "wikiquote.org" } ], [ { "bbox": { "b": 197.2558135986328, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 169.7705078125, "t": 206.21218872070312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "Wikisource" }, { "bbox": { "b": 175.3680877685547, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.6358337402344, "t": 206.09263610839844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "ar, ca, es, eu, fr, id, as, bn, gu, hi, kn, ml, mr, or, pa, ta, te, pt, vi" }, { "bbox": { "b": 197.1860809326172, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 431.0843505859375, "t": 206.09263610839844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "wikisource.org" } ], [ { "bbox": { "b": 162.1378173828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 170.85641479492188, "t": 171.0941925048828 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "Wikiversity" }, { "bbox": { "b": 162.06808471679688, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 336.5514221191406, "t": 170.97463989257812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "ar, en, es, fr, hi, pt, zhs" }, { "bbox": { "b": 162.06808471679688, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 431.1540832519531, "t": 170.97463989257812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "wikiversity.org" } ], [ { "bbox": { "b": 148.8378143310547, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 171.41433715820312, "t": 157.794189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "Wikivoyage" }, { "bbox": { "b": 148.76808166503906, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 351.9037780761719, "t": 157.6746368408203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "en, es, fr, bn, hi, pt, vi, zhs" }, { "bbox": { "b": 148.76808166503906, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 433.3059997558594, "t": 157.6746368408203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "wikivoyage.org" } ], [ { "bbox": { "b": 135.53781127929688, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 170.5077362060547, "t": 144.4941864013672 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "Wiktionary" }, { "bbox": { "b": 124.55906677246094, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.4579772949219, "t": 144.3746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "ar, ca, en, es, eu, fr, id, as, bn, gu, hi, kn, ml, mr, or, pa, ta," }, { "bbox": { "b": 135.46807861328125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 429.8888244628906, "t": 144.3746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "wiktionary.org" } ], [ { "bbox": { "b": 100.4188232421875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 190.00454711914062, "t": 109.37519073486328 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "WuDaoCorpora" }, { "bbox": { "b": 100.34908294677734, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 292.13812255859375, "t": 122.55564880371094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "te, ur, pt, vi zhs" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "" }, { "bbox": { "b": 100.34908294677734, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 443.1590576171875, "t": 109.25565338134766 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "Yuan et al. (2021)" } ], [ { "bbox": { "b": 87.11881256103516, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 169.9697265625, "t": 96.07518768310547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "XQUAD-ca" }, { "bbox": { "b": 87.049072265625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 95.95564270019531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "ca" }, { "bbox": { "b": 76.14007568359375, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 95.95564270019531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "Armengol-Estapé et al. (2021)" } ] ], "num_cols": 3, "num_rows": 27, "table_cells": [ { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 153.51153564453125, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Dataset" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 288.6710510253906, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 707.890869140625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 400.9275817871094, "t": 716.8472290039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Source" }, { "bbox": { "b": 689.9578247070312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 196.01197814941406, "t": 698.9141845703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "The Pile: USPTO" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "en" }, { "bbox": { "b": 689.8880615234375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 439.2835998535156, "t": 698.7946166992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "Gao et al. (2020)" }, { "bbox": { "b": 676.6578369140625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 177.5114288330078, "t": 685.6141967773438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "UIT-VSMEC" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "vi" }, { "bbox": { "b": 676.5880737304688, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 434.8601989746094, "t": 685.49462890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "Ho et al. (2020)" }, { "bbox": { "b": 652.4488525390625, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 220.94834899902344, "t": 672.314208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "United Nations Parallel Corpus" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 311.0870056152344, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "ar, en, es, fr, zhs" }, { "bbox": { "b": 663.2880859375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 455.34332275390625, "t": 672.1946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "Ziemski et al. (2016)" }, { "bbox": { "b": 606.4208984375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 238.77146911621094, "t": 648.105224609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Unsupervised Cross-lingual Representation Learning at Scale Common Crawl Corpus" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 256.0036926269531, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "ne" }, { "bbox": { "b": 639.0791015625, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 458.10296630859375, "t": 647.9856567382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "Conneau et al. (2020)" }, { "bbox": { "b": 593.120849609375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 234.8063201904297, "t": 602.0772094726562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Urdu Monolingual Corpus" }, { "bbox": { "b": 593.0510864257812, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.89784240722656, "t": 601.9576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "ur" }, { "bbox": { "b": 593.0510864257812, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 450.10296630859375, "t": 601.9576416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "Jawaid et al. (2014)" }, { "bbox": { "b": 579.8208618164062, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 170.746826171875, "t": 588.7772216796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "VietAI SAT" }, { "bbox": { "b": 579.7510986328125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 588.6576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "vi" }, { "bbox": { "b": 579.7510986328125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 459.1390686035156, "t": 588.6576538085938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "Ngo and Trinh (2021)" }, { "bbox": { "b": 566.5208740234375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 234.74655151367188, "t": 575.4772338867188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Vietnamese Poetry Corpus" }, { "bbox": { "b": 566.4511108398438, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 575.357666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "vi" }, { "bbox": { "b": 566.4511108398438, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.2275085449219, "t": 575.357666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "Nguyen et al. (2021)" }, { "bbox": { "b": 553.2208251953125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 167.548828125, "t": 562.1771850585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "UIT-VSFC" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "vi" }, { "bbox": { "b": 553.1510620117188, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 454.2275085449219, "t": 562.0576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "Nguyen et al. (2018)" }, { "bbox": { "b": 539.9208374023438, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.44297790527344, "t": 548.877197265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "VilaQuAD" }, { "bbox": { "b": 539.85107421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "ca" }, { "bbox": { "b": 528.9420776367188, "coord_origin": "BOTTOMLEFT", "l": 371.4179992675781, "r": 474.0626525878906, "t": 548.7576293945312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "Rodriguez-Penagos and Armentano-Oller (2021b)" }, { "bbox": { "b": 504.8028259277344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 220.96827697753906, "t": 524.668212890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "VinBigdata-VSLP ASR Challenge 2020" }, { "bbox": { "b": 515.64208984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 524.5486450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "vi" }, { "bbox": { "b": 493.8230895996094, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 536.6311645507812, "t": 524.5486450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "institute.vinbigdata.org/ev ents/vinbigdata- chia-se-100-gio-du-lieu- tieng-noi-cho-cong-dong/" }, { "bbox": { "b": 469.6838073730469, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 240.3455352783203, "t": 489.5491943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "VinBigdata-VSLP Monolin- gual Corpus 2020" }, { "bbox": { "b": 480.5230712890625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 489.42962646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "vi" }, { "bbox": { "b": 458.705078125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 536.6311645507812, "t": 489.42962646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "institute.vinbigdata.org/ev ents/vinbigdata- chia-se-100-gio-du-lieu- tieng-noi-cho-cong-dong/" }, { "bbox": { "b": 434.5658264160156, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 239.80755615234375, "t": 454.43121337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "VinBigdata-VSLP Bilingual Corpus 2020" }, { "bbox": { "b": 445.40509033203125, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 454.3116455078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "vi" }, { "bbox": { "b": 423.58709716796875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 536.6311645507812, "t": 454.3116455078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "institute.vinbigdata.org/ev ents/vinbigdata- chia-se-100-gio-du-lieu- tieng-noi-cho-cong-dong/" }, { "bbox": { "b": 410.3568115234375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 172.5400848388672, "t": 419.3132019042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "ViquiQuAD" }, { "bbox": { "b": 410.2870788574219, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 419.1936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "ca" }, { "bbox": { "b": 399.3780822753906, "coord_origin": "BOTTOMLEFT", "l": 371.4179992675781, "r": 473.5047607421875, "t": 419.1936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "Rodriguez-Penagos and Armentano-Oller (2021c)" }, { "bbox": { "b": 386.1478271484375, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 198.8114776611328, "t": 395.1042175292969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "VNTQcorpus(big)" }, { "bbox": { "b": 386.0780944824219, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 254.3498992919922, "t": 394.9846496582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "vi" }, { "bbox": { "b": 375.1680908203125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 491.60552978515625, "t": 394.9846496582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "http://viet.jnlp.org/download- du-lieu-tu-vung-corpus" }, { "bbox": { "b": 361.93780517578125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 167.2001495361328, "t": 370.8941955566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Wikibooks" }, { "bbox": { "b": 340.0500793457031, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.89910888671875, "t": 370.7746276855469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "ca, ar, en, es, eu, fr, id, bn, hi, ml, mr, pa, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 361.8680725097656, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 428.88262939453125, "t": 370.7746276855469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "wikibooks.org" }, { "bbox": { "b": 326.81982421875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 168.29603576660156, "t": 335.7762145996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Wikimedia" }, { "bbox": { "b": 326.7500915527344, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 293.6424255371094, "t": 335.6566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "ca, id, hi, pt" }, { "bbox": { "b": 326.7500915527344, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 429.4305725097656, "t": 335.6566467285156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "wikimedia.org" }, { "bbox": { "b": 313.5198059082031, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 163.31472778320312, "t": 322.4761962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Wikinews" }, { "bbox": { "b": 313.4500732421875, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 349.8216247558594, "t": 322.35662841796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ar, ca, en, es, fr, ta, pt, zhs" }, { "bbox": { "b": 313.4500732421875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 425.3060607910156, "t": 322.35662841796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "wikinews.org" }, { "bbox": { "b": 300.2198181152344, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 165.53639221191406, "t": 309.17620849609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "Wikipedia" }, { "bbox": { "b": 256.5140686035156, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.2080993652344, "t": 309.056640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "ak, ar, as, bm, bn, ca, en, es, eu, fr, id, ig, gu, hi, ki, kn, lg, ln, ml, mr, nso, ny, or, pa, pt, rn, rw, sn, st, sw, ta, te, tn, ts, tum, tw, ur, vi, wo, yo, zhs," }, { "bbox": { "b": 300.15008544921875, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 426.6609802246094, "t": 309.056640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "wikipedia.org" }, { "bbox": { "b": 232.37380981445312, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 166.08433532714844, "t": 241.33018493652344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "Wikiquote" }, { "bbox": { "b": 210.486083984375, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 365.8990783691406, "t": 241.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "ar, ca, en, es, eu, fr, id, gu, hi, kn, ml, mr, ta, te, ur, pt, vi, zhs" }, { "bbox": { "b": 232.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 427.2188720703125, "t": 241.21063232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "wikiquote.org" }, { "bbox": { "b": 197.2558135986328, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 169.7705078125, "t": 206.21218872070312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "Wikisource" }, { "bbox": { "b": 175.3680877685547, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.6358337402344, "t": 206.09263610839844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "ar, ca, es, eu, fr, id, as, bn, gu, hi, kn, ml, mr, or, pa, ta, te, pt, vi" }, { "bbox": { "b": 197.1860809326172, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 431.0843505859375, "t": 206.09263610839844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "wikisource.org" }, { "bbox": { "b": 162.1378173828125, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 170.85641479492188, "t": 171.0941925048828 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "Wikiversity" }, { "bbox": { "b": 162.06808471679688, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 336.5514221191406, "t": 170.97463989257812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "ar, en, es, fr, hi, pt, zhs" }, { "bbox": { "b": 162.06808471679688, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 431.1540832519531, "t": 170.97463989257812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "wikiversity.org" }, { "bbox": { "b": 148.8378143310547, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 171.41433715820312, "t": 157.794189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "Wikivoyage" }, { "bbox": { "b": 148.76808166503906, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 351.9037780761719, "t": 157.6746368408203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "en, es, fr, bn, hi, pt, vi, zhs" }, { "bbox": { "b": 148.76808166503906, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 433.3059997558594, "t": 157.6746368408203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "wikivoyage.org" }, { "bbox": { "b": 135.53781127929688, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 170.5077362060547, "t": 144.4941864013672 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "Wiktionary" }, { "bbox": { "b": 124.55906677246094, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 366.4579772949219, "t": 144.3746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "ar, ca, en, es, eu, fr, id, as, bn, gu, hi, kn, ml, mr, or, pa, ta," }, { "bbox": { "b": 135.46807861328125, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 429.8888244628906, "t": 144.3746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "wiktionary.org" }, { "bbox": { "b": 100.4188232421875, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 190.00454711914062, "t": 109.37519073486328 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "WuDaoCorpora" }, { "bbox": { "b": 100.34908294677734, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 292.13812255859375, "t": 122.55564880371094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "te, ur, pt, vi zhs" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "" }, { "bbox": { "b": 100.34908294677734, "coord_origin": "BOTTOMLEFT", "l": 371.7770080566406, "r": 443.1590576171875, "t": 109.25565338134766 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "Yuan et al. (2021)" }, { "bbox": { "b": 87.11881256103516, "coord_origin": "BOTTOMLEFT", "l": 121.4219970703125, "r": 169.9697265625, "t": 96.07518768310547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "XQUAD-ca" }, { "bbox": { "b": 87.049072265625, "coord_origin": "BOTTOMLEFT", "l": 246.5989990234375, "r": 255.44578552246094, "t": 95.95564270019531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "ca" }, { "bbox": { "b": 76.14007568359375, "coord_origin": "BOTTOMLEFT", "l": 371.447998046875, "r": 465.6521911621094, "t": 95.95564270019531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "Armengol-Estapé et al. (2021)" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 73.713134765625, "coord_origin": "BOTTOMLEFT", "l": 117.12876892089844, "r": 536.6311645507812, "t": 720.3727416992188 }, "charspan": [ 0, 0 ], "page_no": 30 } ], "references": [], "self_ref": "#/tables/8" }, { "captions": [ { "$ref": "#/texts/339" } ], "children": [], "data": { "grid": [ [ { "bbox": { "b": 646.2171630859375, "coord_origin": "BOTTOMLEFT", "l": 108.2132339477539, "r": 145.9970703125, "t": 655.6136474609375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 646.3136596679688, "coord_origin": "BOTTOMLEFT", "l": 185.93484497070312, "r": 238.85569763183594, "t": 655.48876953125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "ISO-639-3 catalog-ref Genus" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "" }, { "bbox": { "b": 645.921875, "coord_origin": "BOTTOMLEFT", "l": 353.0313720703125, "r": 381.9716491699219, "t": 655.422607421875 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "Family" }, { "bbox": { "b": 645.62890625, "coord_origin": "BOTTOMLEFT", "l": 406.5981140136719, "r": 444.2469787597656, "t": 655.3299560546875 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "Macroarea" }, { "bbox": { "b": 646.5857543945312, "coord_origin": "BOTTOMLEFT", "l": 463.8620300292969, "r": 500.8646545410156, "t": 656.016845703125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "Size in Bytes" } ], [ { "bbox": { "b": 632.6473388671875, "coord_origin": "BOTTOMLEFT", "l": 107.94133758544922, "r": 127.5891342163086, "t": 641.4132080078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Akan" }, { "bbox": { "b": 632.7122802734375, "coord_origin": "BOTTOMLEFT", "l": 204.35984802246094, "r": 216.82046508789062, "t": 641.340576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "aka" }, { "bbox": { "b": 632.6465454101562, "coord_origin": "BOTTOMLEFT", "l": 244.96432495117188, "r": 254.85003662109375, "t": 640.8958740234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "ak" }, { "bbox": { "b": 632.294677734375, "coord_origin": "BOTTOMLEFT", "l": 272.16265869140625, "r": 290.2823791503906, "t": 640.8880615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "Kwa" }, { "bbox": { "b": 632.7719116210938, "coord_origin": "BOTTOMLEFT", "l": 353.5899963378906, "r": 393.4905700683594, "t": 641.7452392578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "Niger-Congo" }, { "bbox": { "b": 632.2013549804688, "coord_origin": "BOTTOMLEFT", "l": 406.6474609375, "r": 427.2049255371094, "t": 640.8191528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "Africa" }, { "bbox": { "b": 633.165283203125, "coord_origin": "BOTTOMLEFT", "l": 470.9044494628906, "r": 500.62353515625, "t": 641.6571044921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "70,1554" } ], [ { "bbox": { "b": 621.754150390625, "coord_origin": "BOTTOMLEFT", "l": 108.563720703125, "r": 133.888671875, "t": 630.2901000976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Arabic" }, { "bbox": { "b": 622.4773559570312, "coord_origin": "BOTTOMLEFT", "l": 205.65953063964844, "r": 215.71127319335938, "t": 630.92822265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "arb" }, { "bbox": { "b": 622.5323486328125, "coord_origin": "BOTTOMLEFT", "l": 246.74658203125, "r": 253.63404846191406, "t": 630.9849243164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "ar" }, { "bbox": { "b": 622.2947387695312, "coord_origin": "BOTTOMLEFT", "l": 272.299072265625, "r": 291.55072021484375, "t": 630.7213134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "Semitic" }, { "bbox": { "b": 621.9561767578125, "coord_origin": "BOTTOMLEFT", "l": 353.9492492675781, "r": 397.2653503417969, "t": 630.767333984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "Afro-Asiatic" }, { "bbox": { "b": 621.6553344726562, "coord_origin": "BOTTOMLEFT", "l": 406.5619201660156, "r": 428.78033447265625, "t": 630.337158203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 2, "text": "Eurasia" }, { "bbox": { "b": 621.7122802734375, "coord_origin": "BOTTOMLEFT", "l": 454.5549011230469, "r": 500.0721435546875, "t": 629.9695434570312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 2, "text": "74,854,900,600" } ], [ { "bbox": { "b": 611.236328125, "coord_origin": "BOTTOMLEFT", "l": 108.90092468261719, "r": 139.33248901367188, "t": 619.7415771484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Assamese" }, { "bbox": { "b": 611.52978515625, "coord_origin": "BOTTOMLEFT", "l": 204.74105834960938, "r": 216.08767700195312, "t": 619.867431640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "asm" }, { "bbox": { "b": 612.06689453125, "coord_origin": "BOTTOMLEFT", "l": 246.48477172851562, "r": 253.59249877929688, "t": 620.5132446289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "as" }, { "bbox": { "b": 611.5089721679688, "coord_origin": "BOTTOMLEFT", "l": 271.9761962890625, "r": 289.5044860839844, "t": 619.9150390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "Indic" }, { "bbox": { "b": 610.7553100585938, "coord_origin": "BOTTOMLEFT", "l": 353.5668640136719, "r": 391.488037109375, "t": 619.4790649414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 611.9335327148438, "coord_origin": "BOTTOMLEFT", "l": 461.8177795410156, "r": 500.3688049316406, "t": 620.1602172851562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 3, "text": "291,522,098" } ], [ { "bbox": { "b": 601.2576293945312, "coord_origin": "BOTTOMLEFT", "l": 108.98766326904297, "r": 137.57203674316406, "t": 609.6809692382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Bambara" }, { "bbox": { "b": 601.77294921875, "coord_origin": "BOTTOMLEFT", "l": 203.88540649414062, "r": 216.4199676513672, "t": 610.0787963867188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "bam" }, { "bbox": { "b": 601.263671875, "coord_origin": "BOTTOMLEFT", "l": 245.21353149414062, "r": 254.23228454589844, "t": 609.674560546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "bm" }, { "bbox": { "b": 601.5784301757812, "coord_origin": "BOTTOMLEFT", "l": 274.1773681640625, "r": 318.1032409667969, "t": 610.1167602539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "Western Mande" }, { "bbox": { "b": 600.5735473632812, "coord_origin": "BOTTOMLEFT", "l": 354.537841796875, "r": 378.57391357421875, "t": 609.1177368164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "Mande" }, { "bbox": { "b": 600.7642211914062, "coord_origin": "BOTTOMLEFT", "l": 406.37811279296875, "r": 426.883544921875, "t": 609.1436157226562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 4, "text": "Africa" }, { "bbox": { "b": 601.4264526367188, "coord_origin": "BOTTOMLEFT", "l": 473.25274658203125, "r": 499.7992858886719, "t": 609.6851806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 4, "text": "391,747" } ], [ { "bbox": { "b": 591.3948974609375, "coord_origin": "BOTTOMLEFT", "l": 108.781494140625, "r": 132.4031982421875, "t": 599.66357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 592.23193359375, "coord_origin": "BOTTOMLEFT", "l": 204.74526977539062, "r": 215.89810180664062, "t": 600.4318237304688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "eus" }, { "bbox": { "b": 591.6302490234375, "coord_origin": "BOTTOMLEFT", "l": 245.8119354248047, "r": 254.22149658203125, "t": 599.9959106445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "eu" }, { "bbox": { "b": 591.2930908203125, "coord_origin": "BOTTOMLEFT", "l": 271.3256530761719, "r": 291.7969970703125, "t": 599.5765380859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 590.8136596679688, "coord_origin": "BOTTOMLEFT", "l": 353.91680908203125, "r": 378.7007751464844, "t": 599.1942749023438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 591.1436157226562, "coord_origin": "BOTTOMLEFT", "l": 406.2830505371094, "r": 427.990966796875, "t": 599.5001220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 5, "text": "Eurasia" }, { "bbox": { "b": 591.21923828125, "coord_origin": "BOTTOMLEFT", "l": 458.5888366699219, "r": 500.11090087890625, "t": 599.5518798828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 5, "text": "2,360,470,848" } ], [ { "bbox": { "b": 580.9913330078125, "coord_origin": "BOTTOMLEFT", "l": 108.32389831542969, "r": 132.4408721923828, "t": 589.3953247070312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Bengali" }, { "bbox": { "b": 581.5091552734375, "coord_origin": "BOTTOMLEFT", "l": 204.27264404296875, "r": 216.28842163085938, "t": 589.8223266601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "ben" }, { "bbox": { "b": 581.3764038085938, "coord_origin": "BOTTOMLEFT", "l": 245.8140106201172, "r": 254.11277770996094, "t": 589.6771240234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "bn" }, { "bbox": { "b": 581.5809326171875, "coord_origin": "BOTTOMLEFT", "l": 270.63018798828125, "r": 287.0025634765625, "t": 589.832763671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 6, "text": "Indic" }, { "bbox": { "b": 580.9334106445312, "coord_origin": "BOTTOMLEFT", "l": 354.4245910644531, "r": 399.6710205078125, "t": 589.3593139648438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 6, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 6, "text": "" }, { "bbox": { "b": 580.7463989257812, "coord_origin": "BOTTOMLEFT", "l": 451.1468505859375, "r": 500.87432861328125, "t": 589.0629272460938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 6, "text": "18,606,823,104" } ], [ { "bbox": { "b": 571.077392578125, "coord_origin": "BOTTOMLEFT", "l": 108.18450164794922, "r": 133.05084228515625, "t": 579.47705078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Catalan" }, { "bbox": { "b": 571.3492431640625, "coord_origin": "BOTTOMLEFT", "l": 205.28199768066406, "r": 215.70187377929688, "t": 579.6483154296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "cat" }, { "bbox": { "b": 571.41259765625, "coord_origin": "BOTTOMLEFT", "l": 246.1710968017578, "r": 253.74913024902344, "t": 579.7540283203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "ca" }, { "bbox": { "b": 571.3926391601562, "coord_origin": "BOTTOMLEFT", "l": 272.9360046386719, "r": 303.5674743652344, "t": 579.6792602539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 7, "text": "Romance" }, { "bbox": { "b": 571.1082763671875, "coord_origin": "BOTTOMLEFT", "l": 354.3467102050781, "r": 399.4031677246094, "t": 579.536376953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 7, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 571.4974365234375, "coord_origin": "BOTTOMLEFT", "l": 451.9903869628906, "r": 500.81103515625, "t": 579.8428955078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 7, "text": "17,792,493,289" } ], [ { "bbox": { "b": 560.6057739257812, "coord_origin": "BOTTOMLEFT", "l": 108.31121826171875, "r": 143.47100830078125, "t": 569.05322265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Chi Chewa" }, { "bbox": { "b": 560.7371826171875, "coord_origin": "BOTTOMLEFT", "l": 204.81883239746094, "r": 216.561279296875, "t": 569.0138549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "nya" }, { "bbox": { "b": 560.5833740234375, "coord_origin": "BOTTOMLEFT", "l": 245.78848266601562, "r": 253.6044921875, "t": 568.9688720703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "ny" }, { "bbox": { "b": 560.8050537109375, "coord_origin": "BOTTOMLEFT", "l": 271.7248840332031, "r": 298.0384216308594, "t": 568.9901733398438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 8, "text": "Bantoid" }, { "bbox": { "b": 560.1963500976562, "coord_origin": "BOTTOMLEFT", "l": 354.5111389160156, "r": 394.291015625, "t": 568.4911499023438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 8, "text": "Niger-Congo" }, { "bbox": { "b": 560.3571166992188, "coord_origin": "BOTTOMLEFT", "l": 406.7816467285156, "r": 427.1576232910156, "t": 568.5674438476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 8, "text": "Africa" }, { "bbox": { "b": 560.4910278320312, "coord_origin": "BOTTOMLEFT", "l": 468.42236328125, "r": 501.07012939453125, "t": 568.7451782226562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 8, "text": "1,187,405" } ], [ { "bbox": { "b": 550.3297729492188, "coord_origin": "BOTTOMLEFT", "l": 108.46611785888672, "r": 142.48959350585938, "t": 558.7322998046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Chi Shona" }, { "bbox": { "b": 550.3215942382812, "coord_origin": "BOTTOMLEFT", "l": 205.2152862548828, "r": 216.55426025390625, "t": 558.6187133789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "sna" }, { "bbox": { "b": 550.130859375, "coord_origin": "BOTTOMLEFT", "l": 246.00352478027344, "r": 253.7533416748047, "t": 558.5576782226562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "sn" }, { "bbox": { "b": 550.825927734375, "coord_origin": "BOTTOMLEFT", "l": 271.8756103515625, "r": 297.1409606933594, "t": 559.0391235351562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 9, "text": "Bantoid" }, { "bbox": { "b": 550.585205078125, "coord_origin": "BOTTOMLEFT", "l": 354.7015075683594, "r": 393.3777160644531, "t": 558.9050903320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 9, "text": "Niger-Congo" }, { "bbox": { "b": 550.6567993164062, "coord_origin": "BOTTOMLEFT", "l": 406.7747802734375, "r": 427.2490539550781, "t": 558.8569946289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 9, "text": "Africa" }, { "bbox": { "b": 550.6953125, "coord_origin": "BOTTOMLEFT", "l": 467.956787109375, "r": 500.91748046875, "t": 558.9978637695312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 9, "text": "6,638,639" } ], [ { "bbox": { "b": 540.658447265625, "coord_origin": "BOTTOMLEFT", "l": 108.03372955322266, "r": 152.3150177001953, "t": 549.0789794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Chi Tumbuka" }, { "bbox": { "b": 540.6922607421875, "coord_origin": "BOTTOMLEFT", "l": 204.47483825683594, "r": 216.28294372558594, "t": 548.9986572265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "tum" }, { "bbox": { "b": 540.871337890625, "coord_origin": "BOTTOMLEFT", "l": 244.4879608154297, "r": 254.66249084472656, "t": 549.2894287109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "tum" }, { "bbox": { "b": 541.38818359375, "coord_origin": "BOTTOMLEFT", "l": 270.99932861328125, "r": 295.212646484375, "t": 549.5797729492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 10, "text": "Bantoid" }, { "bbox": { "b": 540.8161010742188, "coord_origin": "BOTTOMLEFT", "l": 354.9844970703125, "r": 395.4344787597656, "t": 549.1530151367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 10, "text": "Niger-Congo" }, { "bbox": { "b": 540.8040771484375, "coord_origin": "BOTTOMLEFT", "l": 406.79461669921875, "r": 427.968994140625, "t": 549.0009765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 10, "text": "Africa" }, { "bbox": { "b": 540.8871459960938, "coord_origin": "BOTTOMLEFT", "l": 471.609130859375, "r": 501.21917724609375, "t": 549.2110595703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 10, "text": "170,360" } ], [ { "bbox": { "b": 530.3832397460938, "coord_origin": "BOTTOMLEFT", "l": 108.039794921875, "r": 129.19537353515625, "t": 538.6337280273438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "English" }, { "bbox": { "b": 530.1448364257812, "coord_origin": "BOTTOMLEFT", "l": 204.3148193359375, "r": 216.1548614501953, "t": 538.4827880859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "eng" }, { "bbox": { "b": 529.9124145507812, "coord_origin": "BOTTOMLEFT", "l": 244.63050842285156, "r": 254.49581909179688, "t": 538.3505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "en" }, { "bbox": { "b": 530.45556640625, "coord_origin": "BOTTOMLEFT", "l": 272.2205810546875, "r": 301.80712890625, "t": 538.64990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 11, "text": "Germanic" }, { "bbox": { "b": 530.1826171875, "coord_origin": "BOTTOMLEFT", "l": 354.6688537597656, "r": 398.07647705078125, "t": 538.549560546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 11, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 530.55322265625, "coord_origin": "BOTTOMLEFT", "l": 457.59869384765625, "r": 500.9771423339844, "t": 538.8646850585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 11, "text": "484,953,009,124" } ], [ { "bbox": { "b": 520.1136474609375, "coord_origin": "BOTTOMLEFT", "l": 107.9852294921875, "r": 124.80977630615234, "t": 528.241943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Fon" }, { "bbox": { "b": 520.451416015625, "coord_origin": "BOTTOMLEFT", "l": 205.04534912109375, "r": 215.4799346923828, "t": 528.7557983398438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "fon" }, { "bbox": { "b": 520.3941650390625, "coord_origin": "BOTTOMLEFT", "l": 245.62103271484375, "r": 254.02305603027344, "t": 528.8161010742188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "fon" }, { "bbox": { "b": 520.4472045898438, "coord_origin": "BOTTOMLEFT", "l": 272.12701416015625, "r": 287.24969482421875, "t": 528.5364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 12, "text": "Kwa" }, { "bbox": { "b": 520.2186279296875, "coord_origin": "BOTTOMLEFT", "l": 354.25396728515625, "r": 396.1515808105469, "t": 528.5382080078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 12, "text": "Niger-Congo" }, { "bbox": { "b": 520.4370727539062, "coord_origin": "BOTTOMLEFT", "l": 406.46966552734375, "r": 428.5343933105469, "t": 528.7864379882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 12, "text": "Africa" }, { "bbox": { "b": 520.3253173828125, "coord_origin": "BOTTOMLEFT", "l": 464.21484375, "r": 501.8062438964844, "t": 528.512451171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 12, "text": "2,478,546" } ], [ { "bbox": { "b": 510.3376770019531, "coord_origin": "BOTTOMLEFT", "l": 108.0778579711914, "r": 128.46267700195312, "t": 518.528076171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "French" }, { "bbox": { "b": 510.6510314941406, "coord_origin": "BOTTOMLEFT", "l": 204.95750427246094, "r": 215.9420166015625, "t": 519.1000366210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "fra" }, { "bbox": { "b": 510.6396484375, "coord_origin": "BOTTOMLEFT", "l": 246.32420349121094, "r": 253.78955078125, "t": 519.1185302734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "fr" }, { "bbox": { "b": 510.6376953125, "coord_origin": "BOTTOMLEFT", "l": 272.39910888671875, "r": 300.5570373535156, "t": 518.8150634765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 13, "text": "Romance" }, { "bbox": { "b": 510.453369140625, "coord_origin": "BOTTOMLEFT", "l": 354.68988037109375, "r": 400.3841552734375, "t": 518.8603515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 13, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 13, "text": "" } ], [ { "bbox": { "b": 500.30072021484375, "coord_origin": "BOTTOMLEFT", "l": 109.16815185546875, "r": 134.55955505371094, "t": 508.5826721191406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Gujarati" }, { "bbox": { "b": 500.5307312011719, "coord_origin": "BOTTOMLEFT", "l": 205.54954528808594, "r": 215.7245330810547, "t": 508.9515075683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "guj" }, { "bbox": { "b": 500.3470153808594, "coord_origin": "BOTTOMLEFT", "l": 246.97833251953125, "r": 253.26995849609375, "t": 508.82855224609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "gu" }, { "bbox": { "b": 500.1457824707031, "coord_origin": "BOTTOMLEFT", "l": 272.1981201171875, "r": 288.41845703125, "t": 508.32208251953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 14, "text": "Indic" }, { "bbox": { "b": 499.9831848144531, "coord_origin": "BOTTOMLEFT", "l": 354.54937744140625, "r": 401.7975158691406, "t": 508.4107666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 14, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 499.7450256347656, "coord_origin": "BOTTOMLEFT", "l": 452.34222412109375, "r": 501.66302490234375, "t": 516.962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 14, "text": "208,242,620,434 1,199,986,460" } ], [ { "bbox": { "b": 490.5615234375, "coord_origin": "BOTTOMLEFT", "l": 108.0564956665039, "r": 128.54466247558594, "t": 498.7920227050781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Hindi" }, { "bbox": { "b": 490.3254699707031, "coord_origin": "BOTTOMLEFT", "l": 205.20111083984375, "r": 216.18690490722656, "t": 498.85760498046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "hin" }, { "bbox": { "b": 490.57574462890625, "coord_origin": "BOTTOMLEFT", "l": 246.58816528320312, "r": 253.3518524169922, "t": 499.0936279296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "hi" }, { "bbox": { "b": 490.6678771972656, "coord_origin": "BOTTOMLEFT", "l": 271.8505554199219, "r": 289.2873840332031, "t": 498.8848571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 15, "text": "Indic" }, { "bbox": { "b": 490.0447082519531, "coord_origin": "BOTTOMLEFT", "l": 354.6220703125, "r": 400.40203857421875, "t": 498.503662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 15, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 15, "text": "" }, { "bbox": { "b": 490.8396301269531, "coord_origin": "BOTTOMLEFT", "l": 454.7178955078125, "r": 501.1356201171875, "t": 499.1708984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 15, "text": "24,622,119,985" } ], [ { "bbox": { "b": 480.4582824707031, "coord_origin": "BOTTOMLEFT", "l": 108.25355529785156, "r": 123.8819808959961, "t": 488.71112060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Igbo" }, { "bbox": { "b": 480.2000427246094, "coord_origin": "BOTTOMLEFT", "l": 205.50733947753906, "r": 216.15029907226562, "t": 488.73748779296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "ibo" }, { "bbox": { "b": 480.3326416015625, "coord_origin": "BOTTOMLEFT", "l": 246.79440307617188, "r": 253.29974365234375, "t": 488.8615417480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "ig" }, { "bbox": { "b": 480.50146484375, "coord_origin": "BOTTOMLEFT", "l": 271.8854675292969, "r": 291.75390625, "t": 488.7595520019531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 16, "text": "Igboid" }, { "bbox": { "b": 479.901611328125, "coord_origin": "BOTTOMLEFT", "l": 354.8074645996094, "r": 396.034912109375, "t": 488.3544006347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 16, "text": "Niger-Congo" }, { "bbox": { "b": 480.2947998046875, "coord_origin": "BOTTOMLEFT", "l": 406.9546813964844, "r": 427.10858154296875, "t": 488.5445251464844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 16, "text": "Africa" }, { "bbox": { "b": 480.22882080078125, "coord_origin": "BOTTOMLEFT", "l": 467.469482421875, "r": 502.2340393066406, "t": 488.60491943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 16, "text": "14078,521" } ], [ { "bbox": { "b": 470.4344177246094, "coord_origin": "BOTTOMLEFT", "l": 108.68071746826172, "r": 137.19676208496094, "t": 478.73614501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Indonesian" }, { "bbox": { "b": 470.5469970703125, "coord_origin": "BOTTOMLEFT", "l": 204.8577423095703, "r": 216.13925170898438, "t": 479.11810302734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ind" }, { "bbox": { "b": 470.5489501953125, "coord_origin": "BOTTOMLEFT", "l": 246.34152221679688, "r": 253.47520446777344, "t": 479.0879821777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "id" }, { "bbox": { "b": 470.4266357421875, "coord_origin": "BOTTOMLEFT", "l": 273.7782897949219, "r": 325.0870056152344, "t": 478.8460693359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 17, "text": "Malayo-Sumbawan" }, { "bbox": { "b": 470.0202941894531, "coord_origin": "BOTTOMLEFT", "l": 355.7085266113281, "r": 395.0197448730469, "t": 478.4385681152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 17, "text": "Austronesian" }, { "bbox": { "b": 470.5386047363281, "coord_origin": "BOTTOMLEFT", "l": 406.584716796875, "r": 430.518798828125, "t": 478.8246154785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 17, "text": "Papunesia" }, { "bbox": { "b": 470.53857421875, "coord_origin": "BOTTOMLEFT", "l": 457.6726379394531, "r": 501.431884765625, "t": 478.9574279785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 17, "text": "19,972,325,222" } ], [ { "bbox": { "b": 460.09820556640625, "coord_origin": "BOTTOMLEFT", "l": 108.10408020019531, "r": 132.24937438964844, "t": 468.4632568359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "Isi Zulu" }, { "bbox": { "b": 460.4718933105469, "coord_origin": "BOTTOMLEFT", "l": 204.99404907226562, "r": 216.0699005126953, "t": 469.08599853515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "zul" }, { "bbox": { "b": 460.7010498046875, "coord_origin": "BOTTOMLEFT", "l": 246.2052001953125, "r": 253.5150909423828, "t": 469.2560729980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "zu" }, { "bbox": { "b": 461.1459045410156, "coord_origin": "BOTTOMLEFT", "l": 271.419677734375, "r": 299.9014587402344, "t": 469.4657897949219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 18, "text": "Bantoid" }, { "bbox": { "b": 460.0359191894531, "coord_origin": "BOTTOMLEFT", "l": 355.341064453125, "r": 392.1656494140625, "t": 468.4740295410156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 18, "text": "Niger-Congo" }, { "bbox": { "b": 460.8236389160156, "coord_origin": "BOTTOMLEFT", "l": 407.4127502441406, "r": 427.576171875, "t": 469.1214904785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 18, "text": "Africa" }, { "bbox": { "b": 460.66864013671875, "coord_origin": "BOTTOMLEFT", "l": 468.42584228515625, "r": 501.05413818359375, "t": 469.1996765136719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 18, "text": "8,511,561" } ], [ { "bbox": { "b": 450.20770263671875, "coord_origin": "BOTTOMLEFT", "l": 108.04843139648438, "r": 136.48728942871094, "t": 458.585205078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "Kannada" }, { "bbox": { "b": 449.7711181640625, "coord_origin": "BOTTOMLEFT", "l": 204.49652099609375, "r": 216.16058349609375, "t": 458.3982238769531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "kan" }, { "bbox": { "b": 449.9278869628906, "coord_origin": "BOTTOMLEFT", "l": 244.67210388183594, "r": 253.61407470703125, "t": 458.5023498535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "kn" }, { "bbox": { "b": 450.34820556640625, "coord_origin": "BOTTOMLEFT", "l": 272.7766418457031, "r": 336.11004638671875, "t": 458.8428039550781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 19, "text": "Southern Dravidian" }, { "bbox": { "b": 449.8624267578125, "coord_origin": "BOTTOMLEFT", "l": 355.93853759765625, "r": 395.9790344238281, "t": 458.3201904296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 19, "text": "Dravidian" }, { "bbox": { "b": 450.1551513671875, "coord_origin": "BOTTOMLEFT", "l": 406.7572937011719, "r": 429.6174621582031, "t": 458.4588623046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 19, "text": "Eurasia" }, { "bbox": { "b": 450.2169494628906, "coord_origin": "BOTTOMLEFT", "l": 457.0106201171875, "r": 501.13494873046875, "t": 458.651123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 19, "text": "2,098,453,560" } ], [ { "bbox": { "b": 439.8309631347656, "coord_origin": "BOTTOMLEFT", "l": 108.59833526611328, "r": 134.7786865234375, "t": 448.2895812988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "Kikuyu" }, { "bbox": { "b": 439.64404296875, "coord_origin": "BOTTOMLEFT", "l": 204.66603088378906, "r": 216.0258026123047, "t": 448.2994079589844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "kik" }, { "bbox": { "b": 440.22259521484375, "coord_origin": "BOTTOMLEFT", "l": 245.57017517089844, "r": 253.76902770996094, "t": 448.8172912597656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "ki" }, { "bbox": { "b": 440.0103759765625, "coord_origin": "BOTTOMLEFT", "l": 271.54241943359375, "r": 297.26104736328125, "t": 448.4076843261719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 20, "text": "Bantoid" }, { "bbox": { "b": 439.05694580078125, "coord_origin": "BOTTOMLEFT", "l": 354.9132995605469, "r": 394.5080871582031, "t": 447.5910949707031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 20, "text": "Niger-Congo" }, { "bbox": { "b": 440.0498962402344, "coord_origin": "BOTTOMLEFT", "l": 407.20941162109375, "r": 428.545166015625, "t": 448.4403076171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 20, "text": "Africa" }, { "bbox": { "b": 439.6396789550781, "coord_origin": "BOTTOMLEFT", "l": 466.3895568847656, "r": 500.88665771484375, "t": 448.1282043457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 20, "text": "359,615" } ], [ { "bbox": { "b": 429.197509765625, "coord_origin": "BOTTOMLEFT", "l": 108.9115982055664, "r": 154.6288604736328, "t": 437.7027587890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "Kinyarwanda" }, { "bbox": { "b": 429.2398681640625, "coord_origin": "BOTTOMLEFT", "l": 204.80714416503906, "r": 215.96336364746094, "t": 437.84503173828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "kin" }, { "bbox": { "b": 430.0733947753906, "coord_origin": "BOTTOMLEFT", "l": 245.4805145263672, "r": 253.70411682128906, "t": 438.66058349609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "rw" }, { "bbox": { "b": 429.03387451171875, "coord_origin": "BOTTOMLEFT", "l": 271.88140869140625, "r": 296.1385803222656, "t": 437.34600830078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 21, "text": "Bantoid" }, { "bbox": { "b": 429.1513977050781, "coord_origin": "BOTTOMLEFT", "l": 354.9581604003906, "r": 393.7857360839844, "t": 437.66802978515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 21, "text": "Niger-Congo" }, { "bbox": { "b": 429.6646728515625, "coord_origin": "BOTTOMLEFT", "l": 407.10028076171875, "r": 428.4424133300781, "t": 438.06573486328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 21, "text": "Africa" }, { "bbox": { "b": 429.5914001464844, "coord_origin": "BOTTOMLEFT", "l": 466.16143798828125, "r": 500.904052734375, "t": 438.051025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 21, "text": "40,428,299" } ], [ { "bbox": { "b": 418.99176025390625, "coord_origin": "BOTTOMLEFT", "l": 109.14940643310547, "r": 135.82835388183594, "t": 427.58953857421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "Kirundi" }, { "bbox": { "b": 419.1227111816406, "coord_origin": "BOTTOMLEFT", "l": 204.61392211914062, "r": 216.1142578125, "t": 427.8158264160156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "run" }, { "bbox": { "b": 419.64666748046875, "coord_origin": "BOTTOMLEFT", "l": 245.8270263671875, "r": 253.95115661621094, "t": 428.2518005371094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "rn" }, { "bbox": { "b": 419.5191345214844, "coord_origin": "BOTTOMLEFT", "l": 272.36932373046875, "r": 297.0074157714844, "t": 427.9472961425781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 22, "text": "Bantoid" }, { "bbox": { "b": 418.98175048828125, "coord_origin": "BOTTOMLEFT", "l": 354.773681640625, "r": 394.40399169921875, "t": 427.59405517578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 22, "text": "Niger-Congo" }, { "bbox": { "b": 419.12158203125, "coord_origin": "BOTTOMLEFT", "l": 407.1785888671875, "r": 428.7720947265625, "t": 427.5682067871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 22, "text": "Africa" }, { "bbox": { "b": 419.4420166015625, "coord_origin": "BOTTOMLEFT", "l": 467.3368835449219, "r": 500.8492736816406, "t": 427.9061584472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 22, "text": "3,272,550" } ], [ { "bbox": { "b": 409.0594787597656, "coord_origin": "BOTTOMLEFT", "l": 108.30078887939453, "r": 131.46534729003906, "t": 417.6105651855469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "Lingala" }, { "bbox": { "b": 408.3804016113281, "coord_origin": "BOTTOMLEFT", "l": 205.62677001953125, "r": 215.6280517578125, "t": 416.9595947265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "lin" }, { "bbox": { "b": 409.2570495605469, "coord_origin": "BOTTOMLEFT", "l": 246.25428771972656, "r": 253.78628540039062, "t": 417.85589599609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "ln" }, { "bbox": { "b": 408.9028625488281, "coord_origin": "BOTTOMLEFT", "l": 271.39996337890625, "r": 294.73968505859375, "t": 417.22637939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 23, "text": "Bantoid" }, { "bbox": { "b": 409.0890808105469, "coord_origin": "BOTTOMLEFT", "l": 355.3092346191406, "r": 393.3785705566406, "t": 417.5863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 23, "text": "Niger-Congo" }, { "bbox": { "b": 408.7477722167969, "coord_origin": "BOTTOMLEFT", "l": 407.39764404296875, "r": 427.63739013671875, "t": 417.1451110839844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 23, "text": "Africa" }, { "bbox": { "b": 409.3039245605469, "coord_origin": "BOTTOMLEFT", "l": 468.5851745605469, "r": 500.78265380859375, "t": 417.762451171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 23, "text": "1,650,804" } ], [ { "bbox": { "b": 398.44500732421875, "coord_origin": "BOTTOMLEFT", "l": 108.74641418457031, "r": 137.70664978027344, "t": 407.0060119628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "Luganda" }, { "bbox": { "b": 398.3811950683594, "coord_origin": "BOTTOMLEFT", "l": 204.6643524169922, "r": 215.9978790283203, "t": 407.0416564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "lug" }, { "bbox": { "b": 399.0399475097656, "coord_origin": "BOTTOMLEFT", "l": 246.8289794921875, "r": 254.31515502929688, "t": 407.6708984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "lg" }, { "bbox": { "b": 398.84063720703125, "coord_origin": "BOTTOMLEFT", "l": 271.9038391113281, "r": 296.4460754394531, "t": 407.2830810546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 24, "text": "Bantoid" }, { "bbox": { "b": 398.667236328125, "coord_origin": "BOTTOMLEFT", "l": 354.76171875, "r": 394.0806579589844, "t": 407.2316589355469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 24, "text": "Niger-Congo" }, { "bbox": { "b": 398.8894348144531, "coord_origin": "BOTTOMLEFT", "l": 407.5158996582031, "r": 427.80133056640625, "t": 407.3270263671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 24, "text": "Africa" }, { "bbox": { "b": 399.0642395019531, "coord_origin": "BOTTOMLEFT", "l": 466.9955749511719, "r": 501.0726623535156, "t": 407.54583740234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 24, "text": "4,568,367" } ], [ { "bbox": { "b": 388.86865234375, "coord_origin": "BOTTOMLEFT", "l": 108.54914855957031, "r": 141.93531799316406, "t": 397.4731140136719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "Malayalam" }, { "bbox": { "b": 388.5834045410156, "coord_origin": "BOTTOMLEFT", "l": 204.49880981445312, "r": 216.2100067138672, "t": 397.2698059082031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "mal" }, { "bbox": { "b": 388.9284973144531, "coord_origin": "BOTTOMLEFT", "l": 245.669677734375, "r": 253.9735870361328, "t": 397.54095458984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "ml" }, { "bbox": { "b": 388.9745178222656, "coord_origin": "BOTTOMLEFT", "l": 271.04913330078125, "r": 310.7453918457031, "t": 397.46856689453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 25, "text": "Southern Dravidian" }, { "bbox": { "b": 388.6893310546875, "coord_origin": "BOTTOMLEFT", "l": 355.494140625, "r": 394.22955322265625, "t": 397.2818908691406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 25, "text": "Dravidian" }, { "bbox": { "b": 388.61322021484375, "coord_origin": "BOTTOMLEFT", "l": 407.1199035644531, "r": 429.5668029785156, "t": 397.1412353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 25, "text": "Eurasia" }, { "bbox": { "b": 388.4208679199219, "coord_origin": "BOTTOMLEFT", "l": 458.1934509277344, "r": 501.4151306152344, "t": 396.88128662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 25, "text": "3,662,571,498" } ], [ { "bbox": { "b": 378.7137756347656, "coord_origin": "BOTTOMLEFT", "l": 109.17071533203125, "r": 135.9722442626953, "t": 387.33428955078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "Marathi" }, { "bbox": { "b": 377.8516845703125, "coord_origin": "BOTTOMLEFT", "l": 204.9829864501953, "r": 216.25836181640625, "t": 386.425048828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "mar" }, { "bbox": { "b": 378.8268737792969, "coord_origin": "BOTTOMLEFT", "l": 245.8630828857422, "r": 253.68673706054688, "t": 387.4132385253906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "mr" }, { "bbox": { "b": 378.7937927246094, "coord_origin": "BOTTOMLEFT", "l": 270.7461242675781, "r": 292.65325927734375, "t": 387.2211608886719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 26, "text": "Indic" }, { "bbox": { "b": 378.4562683105469, "coord_origin": "BOTTOMLEFT", "l": 355.687255859375, "r": 396.0417785644531, "t": 387.03192138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 26, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 26, "text": "" }, { "bbox": { "b": 377.6427917480469, "coord_origin": "BOTTOMLEFT", "l": 454.1509094238281, "r": 501.5225830078125, "t": 386.1326904296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 26, "text": "1,775,483,122" } ], [ { "bbox": { "b": 368.8480529785156, "coord_origin": "BOTTOMLEFT", "l": 108.32581329345703, "r": 128.98614501953125, "t": 377.3013916015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 27, "text": "Nepali" }, { "bbox": { "b": 368.4093322753906, "coord_origin": "BOTTOMLEFT", "l": 205.1424102783203, "r": 216.29241943359375, "t": 377.0423889160156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 27, "text": "nep" }, { "bbox": { "b": 368.9896240234375, "coord_origin": "BOTTOMLEFT", "l": 245.89109802246094, "r": 253.8175811767578, "t": 377.5635681152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 27, "text": "ne" }, { "bbox": { "b": 369.712890625, "coord_origin": "BOTTOMLEFT", "l": 270.24749755859375, "r": 293.37432861328125, "t": 378.1374816894531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 27, "text": "Indic" }, { "bbox": { "b": 368.70111083984375, "coord_origin": "BOTTOMLEFT", "l": 355.7784118652344, "r": 398.30487060546875, "t": 377.30120849609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 27, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 27, "text": "" }, { "bbox": { "b": 368.2612609863281, "coord_origin": "BOTTOMLEFT", "l": 455.61175537109375, "r": 501.53887939453125, "t": 376.69549560546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 27, "text": "2,551,307,393" } ], [ { "bbox": { "b": 357.67095947265625, "coord_origin": "BOTTOMLEFT", "l": 109.35924530029297, "r": 155.69496154785156, "t": 366.2355041503906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 28, "text": "Northern Sotho" }, { "bbox": { "b": 358.3829650878906, "coord_origin": "BOTTOMLEFT", "l": 205.17263793945312, "r": 216.2655029296875, "t": 366.9328308105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 28, "text": "nso" }, { "bbox": { "b": 358.58172607421875, "coord_origin": "BOTTOMLEFT", "l": 245.02195739746094, "r": 253.70367431640625, "t": 367.1554260253906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 28, "text": "nso" }, { "bbox": { "b": 358.9200134277344, "coord_origin": "BOTTOMLEFT", "l": 271.0533142089844, "r": 293.208740234375, "t": 367.3212585449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 28, "text": "Bantoid" }, { "bbox": { "b": 358.2391052246094, "coord_origin": "BOTTOMLEFT", "l": 355.3683776855469, "r": 396.8522644042969, "t": 366.7705993652344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 28, "text": "Niger-Congo" }, { "bbox": { "b": 358.4185791015625, "coord_origin": "BOTTOMLEFT", "l": 407.396728515625, "r": 428.9354553222656, "t": 366.8617858886719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 28, "text": "Africa" }, { "bbox": { "b": 358.656005859375, "coord_origin": "BOTTOMLEFT", "l": 467.3802490234375, "r": 500.82958984375, "t": 367.0387878417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 28, "text": "1,764,506" } ], [ { "bbox": { "b": 348.6308898925781, "coord_origin": "BOTTOMLEFT", "l": 108.28306579589844, "r": 126.74258422851562, "t": 357.0397644042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 29, "text": "Odia" }, { "bbox": { "b": 348.3537902832031, "coord_origin": "BOTTOMLEFT", "l": 204.96728515625, "r": 216.32981872558594, "t": 356.9362487792969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 29, "text": "ori" }, { "bbox": { "b": 349.1275939941406, "coord_origin": "BOTTOMLEFT", "l": 245.6078338623047, "r": 253.9015655517578, "t": 357.6670227050781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 29, "text": "or" }, { "bbox": { "b": 349.31854248046875, "coord_origin": "BOTTOMLEFT", "l": 270.2909240722656, "r": 291.72900390625, "t": 357.7449035644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 29, "text": "Indic" }, { "bbox": { "b": 348.29119873046875, "coord_origin": "BOTTOMLEFT", "l": 355.2393798828125, "r": 400.64923095703125, "t": 356.9054870605469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 29, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 29, "text": "" }, { "bbox": { "b": 348.8809509277344, "coord_origin": "BOTTOMLEFT", "l": 455.0662841796875, "r": 501.3878173828125, "t": 357.31317138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 29, "text": "1,157,100,133" } ], [ { "bbox": { "b": 338.6387634277344, "coord_origin": "BOTTOMLEFT", "l": 108.17900848388672, "r": 140.20892333984375, "t": 347.0970458984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 30, "text": "Portuguese" }, { "bbox": { "b": 338.3876953125, "coord_origin": "BOTTOMLEFT", "l": 205.31768798828125, "r": 216.0242156982422, "t": 346.7798767089844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 30, "text": "por" }, { "bbox": { "b": 338.287353515625, "coord_origin": "BOTTOMLEFT", "l": 246.67745971679688, "r": 253.61895751953125, "t": 346.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 30, "text": "pt" }, { "bbox": { "b": 339.12713623046875, "coord_origin": "BOTTOMLEFT", "l": 270.8980407714844, "r": 298.5016174316406, "t": 347.4864501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 30, "text": "Romance" }, { "bbox": { "b": 337.8818664550781, "coord_origin": "BOTTOMLEFT", "l": 355.2070007324219, "r": 401.1482238769531, "t": 346.43310546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 30, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 30, "text": "" }, { "bbox": { "b": 338.24639892578125, "coord_origin": "BOTTOMLEFT", "l": 453.4613952636719, "r": 501.1967468261719, "t": 346.6802062988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 30, "text": "79,277,543,375" } ], [ { "bbox": { "b": 327.7706604003906, "coord_origin": "BOTTOMLEFT", "l": 108.52886962890625, "r": 132.45672607421875, "t": 336.1958312988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 31, "text": "Punjabi" }, { "bbox": { "b": 328.3049621582031, "coord_origin": "BOTTOMLEFT", "l": 205.0506134033203, "r": 216.06906127929688, "t": 336.6721496582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 31, "text": "pan" }, { "bbox": { "b": 328.44378662109375, "coord_origin": "BOTTOMLEFT", "l": 246.59326171875, "r": 253.76263427734375, "t": 336.8516845703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 31, "text": "pa" }, { "bbox": { "b": 328.44622802734375, "coord_origin": "BOTTOMLEFT", "l": 271.5168762207031, "r": 286.8114929199219, "t": 336.776611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 31, "text": "Indic" }, { "bbox": { "b": 327.73773193359375, "coord_origin": "BOTTOMLEFT", "l": 355.3795166015625, "r": 399.90777587890625, "t": 336.2171325683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 31, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 31, "text": "" }, { "bbox": { "b": 328.4530334472656, "coord_origin": "BOTTOMLEFT", "l": 462.5596008300781, "r": 500.4649963378906, "t": 336.83160400390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 31, "text": "1,572,109,752" } ], [ { "bbox": { "b": 317.9649658203125, "coord_origin": "BOTTOMLEFT", "l": 108.49153900146484, "r": 134.1531524658203, "t": 326.4454345703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 32, "text": "Sesotho" }, { "bbox": { "b": 317.9165344238281, "coord_origin": "BOTTOMLEFT", "l": 205.4767608642578, "r": 215.55690002441406, "t": 326.3720397949219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 32, "text": "sot" }, { "bbox": { "b": 318.0326843261719, "coord_origin": "BOTTOMLEFT", "l": 246.9828338623047, "r": 253.73391723632812, "t": 326.5728759765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 32, "text": "st" }, { "bbox": { "b": 318.3270263671875, "coord_origin": "BOTTOMLEFT", "l": 272.4049072265625, "r": 295.37548828125, "t": 326.66546630859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 32, "text": "Bantoid" }, { "bbox": { "b": 317.82415771484375, "coord_origin": "BOTTOMLEFT", "l": 354.586181640625, "r": 396.1689758300781, "t": 326.36676025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 32, "text": "Niger-Congo" }, { "bbox": { "b": 318.34857177734375, "coord_origin": "BOTTOMLEFT", "l": 407.21234130859375, "r": 427.7130432128906, "t": 326.7799072265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 32, "text": "Africa" }, { "bbox": { "b": 318.2388916015625, "coord_origin": "BOTTOMLEFT", "l": 473.3670654296875, "r": 500.52459716796875, "t": 326.6830749511719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 32, "text": "751,034" } ], [ { "bbox": { "b": 307.5071105957031, "coord_origin": "BOTTOMLEFT", "l": 108.30636596679688, "r": 136.1265411376953, "t": 315.8783264160156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 33, "text": "Setswana" }, { "bbox": { "b": 307.56317138671875, "coord_origin": "BOTTOMLEFT", "l": 205.66221618652344, "r": 215.0456085205078, "t": 316.0390930175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 33, "text": "tsn" }, { "bbox": { "b": 308.10260009765625, "coord_origin": "BOTTOMLEFT", "l": 247.04190063476562, "r": 253.63511657714844, "t": 316.6589660644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 33, "text": "tn" }, { "bbox": { "b": 307.83221435546875, "coord_origin": "BOTTOMLEFT", "l": 273.2046813964844, "r": 296.1052551269531, "t": 316.2181701660156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 33, "text": "Bantoid" }, { "bbox": { "b": 307.2530822753906, "coord_origin": "BOTTOMLEFT", "l": 353.7736511230469, "r": 392.70672607421875, "t": 315.7368469238281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 33, "text": "Niger-Congo" }, { "bbox": { "b": 307.73291015625, "coord_origin": "BOTTOMLEFT", "l": 407.1366271972656, "r": 427.6194152832031, "t": 316.18487548828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 33, "text": "Africa" }, { "bbox": { "b": 308.2646484375, "coord_origin": "BOTTOMLEFT", "l": 471.7481384277344, "r": 500.15313720703125, "t": 316.6508483886719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 33, "text": "1,502,200" } ], [ { "bbox": { "b": 297.7591247558594, "coord_origin": "BOTTOMLEFT", "l": 108.14258575439453, "r": 169.5782928466797, "t": 306.10028076171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 34, "text": "Simplified Chinese" }, { "bbox": { "b": 297.88739013671875, "coord_origin": "BOTTOMLEFT", "l": 205.19297790527344, "r": 215.63514709472656, "t": 306.369384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 34, "text": "-" }, { "bbox": { "b": 298.06085205078125, "coord_origin": "BOTTOMLEFT", "l": 244.327392578125, "r": 254.46121215820312, "t": 306.5933532714844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 34, "text": "zhs" }, { "bbox": { "b": 298.551513671875, "coord_origin": "BOTTOMLEFT", "l": 271.9973449707031, "r": 298.54327392578125, "t": 306.8904724121094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 34, "text": "Chinese" }, { "bbox": { "b": 297.8870849609375, "coord_origin": "BOTTOMLEFT", "l": 354.0538024902344, "r": 396.21484375, "t": 306.44537353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 34, "text": "Sino-Tibetan" }, { "bbox": { "b": 297.8769226074219, "coord_origin": "BOTTOMLEFT", "l": 406.57568359375, "r": 429.7794189453125, "t": 306.42022705078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 34, "text": "Eurasia" }, { "bbox": { "b": 297.3220520019531, "coord_origin": "BOTTOMLEFT", "l": 450.81640625, "r": 500.7181396484375, "t": 305.8003234863281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 34, "text": "261,019,433,892" } ], [ { "bbox": { "b": 287.2420654296875, "coord_origin": "BOTTOMLEFT", "l": 108.35747528076172, "r": 134.9157257080078, "t": 295.62994384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 35, "text": "Spanish" }, { "bbox": { "b": 287.7139587402344, "coord_origin": "BOTTOMLEFT", "l": 205.1435089111328, "r": 215.99790954589844, "t": 296.11444091796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 35, "text": "spa" }, { "bbox": { "b": 287.77813720703125, "coord_origin": "BOTTOMLEFT", "l": 245.9720458984375, "r": 254.34469604492188, "t": 296.24041748046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 35, "text": "es" }, { "bbox": { "b": 287.5814208984375, "coord_origin": "BOTTOMLEFT", "l": 271.341796875, "r": 301.1479797363281, "t": 295.8623352050781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 35, "text": "Romance" }, { "bbox": { "b": 287.2391357421875, "coord_origin": "BOTTOMLEFT", "l": 354.4414367675781, "r": 399.9187927246094, "t": 295.7132873535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 35, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 35, "text": "" }, { "bbox": { "b": 287.7720031738281, "coord_origin": "BOTTOMLEFT", "l": 448.2947692871094, "r": 500.7635192871094, "t": 296.0963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 35, "text": "175,098,365,045" } ], [ { "bbox": { "b": 277.56463623046875, "coord_origin": "BOTTOMLEFT", "l": 108.71043395996094, "r": 132.465576171875, "t": 286.0724792480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 36, "text": "Swahili" }, { "bbox": { "b": 277.816162109375, "coord_origin": "BOTTOMLEFT", "l": 204.20738220214844, "r": 216.68763732910156, "t": 286.2982177734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 36, "text": "swh" }, { "bbox": { "b": 278.0283203125, "coord_origin": "BOTTOMLEFT", "l": 245.92393493652344, "r": 253.94558715820312, "t": 286.57135009765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 36, "text": "sw" }, { "bbox": { "b": 278.00048828125, "coord_origin": "BOTTOMLEFT", "l": 271.4209899902344, "r": 300.7360534667969, "t": 286.3871154785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 36, "text": "Bantoid" }, { "bbox": { "b": 277.2432861328125, "coord_origin": "BOTTOMLEFT", "l": 354.04034423828125, "r": 393.48223876953125, "t": 285.7327880859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 36, "text": "Niger-Congo" }, { "bbox": { "b": 277.5732727050781, "coord_origin": "BOTTOMLEFT", "l": 406.56268310546875, "r": 429.0010986328125, "t": 286.0862731933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 36, "text": "Africa" }, { "bbox": { "b": 277.6167907714844, "coord_origin": "BOTTOMLEFT", "l": 456.8694763183594, "r": 501.2992248535156, "t": 285.93365478515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 36, "text": "236,482,543" } ], [ { "bbox": { "b": 266.7707214355469, "coord_origin": "BOTTOMLEFT", "l": 108.87135314941406, "r": 130.01206970214844, "t": 275.24664306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 37, "text": "Tamil" }, { "bbox": { "b": 267.4174499511719, "coord_origin": "BOTTOMLEFT", "l": 203.4427032470703, "r": 217.20338439941406, "t": 275.8340148925781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 37, "text": "tam" }, { "bbox": { "b": 267.2631530761719, "coord_origin": "BOTTOMLEFT", "l": 246.16098022460938, "r": 253.42066955566406, "t": 275.7876281738281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 37, "text": "ta" }, { "bbox": { "b": 266.78253173828125, "coord_origin": "BOTTOMLEFT", "l": 273.2210693359375, "r": 339.6412353515625, "t": 275.2291259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 37, "text": "Southern Dravidian" }, { "bbox": { "b": 266.7969665527344, "coord_origin": "BOTTOMLEFT", "l": 354.81964111328125, "r": 383.95281982421875, "t": 275.200439453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 37, "text": "Dravidian" }, { "bbox": { "b": 267.22747802734375, "coord_origin": "BOTTOMLEFT", "l": 406.36419677734375, "r": 429.8748474121094, "t": 275.7294616699219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 37, "text": "Eurasia" }, { "bbox": { "b": 267.181640625, "coord_origin": "BOTTOMLEFT", "l": 456.1546936035156, "r": 501.1959533691406, "t": 275.45147705078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 37, "text": "7,989,206,220" } ], [ { "bbox": { "b": 257.2725524902344, "coord_origin": "BOTTOMLEFT", "l": 108.70472717285156, "r": 128.7847442626953, "t": 265.7151794433594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 38, "text": "Telugu" }, { "bbox": { "b": 256.9312438964844, "coord_origin": "BOTTOMLEFT", "l": 204.9400634765625, "r": 215.74571228027344, "t": 265.36444091796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 38, "text": "tel" }, { "bbox": { "b": 257.34429931640625, "coord_origin": "BOTTOMLEFT", "l": 246.07130432128906, "r": 253.7733154296875, "t": 265.8937683105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 38, "text": "te" }, { "bbox": { "b": 257.2600402832031, "coord_origin": "BOTTOMLEFT", "l": 273.256103515625, "r": 346.8114929199219, "t": 265.7115478515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 38, "text": "South-Central Dravidian Dravidian" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 38, "text": "" }, { "bbox": { "b": 257.0427551269531, "coord_origin": "BOTTOMLEFT", "l": 406.3435974121094, "r": 429.6686096191406, "t": 265.54815673828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 38, "text": "Eurasia" }, { "bbox": { "b": 257.5414733886719, "coord_origin": "BOTTOMLEFT", "l": 460.3102111816406, "r": 501.2451171875, "t": 265.8232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 38, "text": "2993407,159" } ], [ { "bbox": { "b": 246.57666015625, "coord_origin": "BOTTOMLEFT", "l": 108.58483123779297, "r": 171.8043975830078, "t": 254.88795471191406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 39, "text": "Traditional Chinese" }, { "bbox": { "b": 246.85757446289062, "coord_origin": "BOTTOMLEFT", "l": 205.56141662597656, "r": 214.78268432617188, "t": 255.35275268554688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 39, "text": "-" }, { "bbox": { "b": 246.96266174316406, "coord_origin": "BOTTOMLEFT", "l": 245.1876983642578, "r": 254.34671020507812, "t": 255.5424041748047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 39, "text": "zht" }, { "bbox": { "b": 247.37074279785156, "coord_origin": "BOTTOMLEFT", "l": 272.2436828613281, "r": 303.81634521484375, "t": 255.70603942871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 39, "text": "Chinese" }, { "bbox": { "b": 247.2266082763672, "coord_origin": "BOTTOMLEFT", "l": 355.0390625, "r": 393.97882080078125, "t": 255.6996307373047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 39, "text": "Sino-Tibetan" }, { "bbox": { "b": 246.9362335205078, "coord_origin": "BOTTOMLEFT", "l": 406.3245849609375, "r": 429.05096435546875, "t": 255.42367553710938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 39, "text": "Eurasia" }, { "bbox": { "b": 246.86868286132812, "coord_origin": "BOTTOMLEFT", "l": 463.15142822265625, "r": 501.22027587890625, "t": 255.15745544433594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 39, "text": "762,489,150" } ], [ { "bbox": { "b": 236.9476318359375, "coord_origin": "BOTTOMLEFT", "l": 108.84326934814453, "r": 123.63748168945312, "t": 245.3838653564453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 40, "text": "Twi" }, { "bbox": { "b": 236.7466583251953, "coord_origin": "BOTTOMLEFT", "l": 205.5863037109375, "r": 215.42103576660156, "t": 245.21051025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 40, "text": "twi" }, { "bbox": { "b": 237.1617889404297, "coord_origin": "BOTTOMLEFT", "l": 246.14259338378906, "r": 253.93890380859375, "t": 245.7530059814453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 40, "text": "tw" }, { "bbox": { "b": 237.04428100585938, "coord_origin": "BOTTOMLEFT", "l": 271.3865966796875, "r": 286.4140319824219, "t": 245.48175048828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 40, "text": "Kwa" }, { "bbox": { "b": 236.35682678222656, "coord_origin": "BOTTOMLEFT", "l": 354.77606201171875, "r": 398.3659973144531, "t": 244.86277770996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 40, "text": "Niger-Congo" }, { "bbox": { "b": 236.8165740966797, "coord_origin": "BOTTOMLEFT", "l": 406.5733947753906, "r": 428.7755432128906, "t": 245.31649780273438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 40, "text": "Africa" }, { "bbox": { "b": 236.84291076660156, "coord_origin": "BOTTOMLEFT", "l": 467.62652587890625, "r": 500.6257019042969, "t": 245.17544555664062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 40, "text": "1,265,041" } ], [ { "bbox": { "b": 227.35032653808594, "coord_origin": "BOTTOMLEFT", "l": 108.52404022216797, "r": 124.53902435302734, "t": 235.76144409179688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 41, "text": "Urdu" }, { "bbox": { "b": 227.0521240234375, "coord_origin": "BOTTOMLEFT", "l": 205.2458953857422, "r": 216.35084533691406, "t": 235.49473571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 41, "text": "urd" }, { "bbox": { "b": 227.05445861816406, "coord_origin": "BOTTOMLEFT", "l": 246.3316192626953, "r": 253.7119903564453, "t": 235.62001037597656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 41, "text": "ur" }, { "bbox": { "b": 227.7138214111328, "coord_origin": "BOTTOMLEFT", "l": 271.7252197265625, "r": 284.8417053222656, "t": 236.22555541992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 41, "text": "Indic" }, { "bbox": { "b": 226.95724487304688, "coord_origin": "BOTTOMLEFT", "l": 354.75848388671875, "r": 402.03424072265625, "t": 235.4998016357422 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 41, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 41, "text": "" }, { "bbox": { "b": 227.02642822265625, "coord_origin": "BOTTOMLEFT", "l": 454.41302490234375, "r": 500.4913635253906, "t": 235.41854858398438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 41, "text": "2,781,329,959" } ], [ { "bbox": { "b": 216.3731689453125, "coord_origin": "BOTTOMLEFT", "l": 108.97728729248047, "r": 144.2902069091797, "t": 224.86988830566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 42, "text": "Vietnamese" }, { "bbox": { "b": 216.7290802001953, "coord_origin": "BOTTOMLEFT", "l": 205.62844848632812, "r": 216.28665161132812, "t": 225.19895935058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 42, "text": "vie" }, { "bbox": { "b": 216.7925567626953, "coord_origin": "BOTTOMLEFT", "l": 246.4274139404297, "r": 253.82920837402344, "t": 225.45684814453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 42, "text": "vi" }, { "bbox": { "b": 217.48045349121094, "coord_origin": "BOTTOMLEFT", "l": 271.5227966308594, "r": 308.4107971191406, "t": 225.9589080810547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 42, "text": "Viet-Muong" }, { "bbox": { "b": 216.5958709716797, "coord_origin": "BOTTOMLEFT", "l": 354.7583923339844, "r": 401.1172180175781, "t": 225.18240356445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 42, "text": "Austro-Asiatic" }, { "bbox": { "b": 216.71151733398438, "coord_origin": "BOTTOMLEFT", "l": 406.1020812988281, "r": 429.02044677734375, "t": 225.3156280517578 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 42, "text": "Eurasia" }, { "bbox": { "b": 217.2748565673828, "coord_origin": "BOTTOMLEFT", "l": 456.14923095703125, "r": 500.35308837890625, "t": 225.66273498535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 42, "text": "43,709,279,959" } ], [ { "bbox": { "b": 206.9718475341797, "coord_origin": "BOTTOMLEFT", "l": 108.85558319091797, "r": 129.42886352539062, "t": 215.3903350830078 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 43, "text": "Wolof" }, { "bbox": { "b": 206.7169647216797, "coord_origin": "BOTTOMLEFT", "l": 205.2184295654297, "r": 216.8064422607422, "t": 215.29913330078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 43, "text": "wol" }, { "bbox": { "b": 206.982421875, "coord_origin": "BOTTOMLEFT", "l": 245.613037109375, "r": 254.1932830810547, "t": 215.71926879882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 43, "text": "wo" }, { "bbox": { "b": 207.22157287597656, "coord_origin": "BOTTOMLEFT", "l": 270.56488037109375, "r": 290.053466796875, "t": 215.87628173828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 43, "text": "Wolof" }, { "bbox": { "b": 206.91802978515625, "coord_origin": "BOTTOMLEFT", "l": 354.80926513671875, "r": 398.0904235839844, "t": 215.5409393310547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 43, "text": "Niger-Congo" }, { "bbox": { "b": 206.661865234375, "coord_origin": "BOTTOMLEFT", "l": 406.0134582519531, "r": 428.2970886230469, "t": 215.3527069091797 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 43, "text": "Africa" }, { "bbox": { "b": 206.67898559570312, "coord_origin": "BOTTOMLEFT", "l": 466.620849609375, "r": 500.9853820800781, "t": 215.0123291015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 43, "text": "3,606,973" } ], [ { "bbox": { "b": 196.03115844726562, "coord_origin": "BOTTOMLEFT", "l": 108.87499237060547, "r": 130.56085205078125, "t": 204.55064392089844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 44, "text": "Xhosa" }, { "bbox": { "b": 196.32159423828125, "coord_origin": "BOTTOMLEFT", "l": 204.9607696533203, "r": 216.56484985351562, "t": 204.9009246826172 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 44, "text": "xho" }, { "bbox": { "b": 196.44464111328125, "coord_origin": "BOTTOMLEFT", "l": 246.29147338867188, "r": 253.8073272705078, "t": 205.17361450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 44, "text": "xh" }, { "bbox": { "b": 195.97079467773438, "coord_origin": "BOTTOMLEFT", "l": 271.2140197753906, "r": 291.8352355957031, "t": 204.65423583984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 44, "text": "Bantoid" }, { "bbox": { "b": 195.67611694335938, "coord_origin": "BOTTOMLEFT", "l": 355.1220703125, "r": 396.5032043457031, "t": 204.26348876953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 44, "text": "Niger-Congo" }, { "bbox": { "b": 195.8412628173828, "coord_origin": "BOTTOMLEFT", "l": 405.8934020996094, "r": 427.184814453125, "t": 204.533935546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 44, "text": "Africa" }, { "bbox": { "b": 195.99139404296875, "coord_origin": "BOTTOMLEFT", "l": 466.9107971191406, "r": 500.642822265625, "t": 204.3134002685547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 44, "text": "14,304,074" } ], [ { "bbox": { "b": 185.78851318359375, "coord_origin": "BOTTOMLEFT", "l": 108.67925262451172, "r": 134.6645050048828, "t": 194.46615600585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 45, "text": "Xitsonga" }, { "bbox": { "b": 186.1658935546875, "coord_origin": "BOTTOMLEFT", "l": 205.25071716308594, "r": 215.73153686523438, "t": 194.8161163330078 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 45, "text": "tso" }, { "bbox": { "b": 186.6497039794922, "coord_origin": "BOTTOMLEFT", "l": 246.6892852783203, "r": 253.50070190429688, "t": 195.4151611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 45, "text": "ts" }, { "bbox": { "b": 186.28427124023438, "coord_origin": "BOTTOMLEFT", "l": 272.11260986328125, "r": 294.1134338378906, "t": 194.98060607910156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 45, "text": "Bantoid" }, { "bbox": { "b": 186.032470703125, "coord_origin": "BOTTOMLEFT", "l": 355.0280456542969, "r": 396.8128967285156, "t": 194.68316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 45, "text": "Niger-Congo" }, { "bbox": { "b": 186.06903076171875, "coord_origin": "BOTTOMLEFT", "l": 406.20843505859375, "r": 426.8867492675781, "t": 194.7771453857422 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 45, "text": "Africa" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 45, "text": "" } ], [ { "bbox": { "b": 175.14283752441406, "coord_origin": "BOTTOMLEFT", "l": 108.71174621582031, "r": 139.44134521484375, "t": 183.88453674316406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 46, "text": "Yoruba" }, { "bbox": { "b": 174.76832580566406, "coord_origin": "BOTTOMLEFT", "l": 205.20335388183594, "r": 214.9654998779297, "t": 183.44154357910156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 46, "text": "yor" }, { "bbox": { "b": 175.76734924316406, "coord_origin": "BOTTOMLEFT", "l": 246.59974670410156, "r": 253.18612670898438, "t": 184.48805236816406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 46, "text": "yo" }, { "bbox": { "b": 175.52835083007812, "coord_origin": "BOTTOMLEFT", "l": 272.08477783203125, "r": 293.53045654296875, "t": 184.17129516601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 46, "text": "Defoid" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 46, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 46, "text": "" }, { "bbox": { "b": 183.8022003173828, "coord_origin": "BOTTOMLEFT", "l": 469.1238708496094, "r": 500.5533142089844, "t": 192.45419311523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 46, "text": "707,634" } ], [ { "bbox": { "b": 163.75592041015625, "coord_origin": "BOTTOMLEFT", "l": 108.24192810058594, "r": 187.58399963378906, "t": 172.73565673828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 47, "text": "Programming Languages" }, { "bbox": { "b": 164.87973022460938, "coord_origin": "BOTTOMLEFT", "l": 204.60964965820312, "r": 214.92276000976562, "t": 173.8241729736328 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 47, "text": "-" }, { "bbox": { "b": 165.4375457763672, "coord_origin": "BOTTOMLEFT", "l": 246.6812286376953, "r": 253.26661682128906, "t": 174.3863525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 47, "text": "-" }, { "bbox": { "b": 167.28050231933594, "coord_origin": "BOTTOMLEFT", "l": 271.4575500488281, "r": 291.3991394042969, "t": 176.1390838623047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 47, "text": "-" }, { "bbox": { "b": 164.84762573242188, "coord_origin": "BOTTOMLEFT", "l": 355.00469970703125, "r": 395.3666687011719, "t": 183.07347106933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 47, "text": "Niger-Congo -" }, { "bbox": { "b": 174.52589416503906, "coord_origin": "BOTTOMLEFT", "l": 406.2784423828125, "r": 426.1576843261719, "t": 183.22804260253906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 47, "text": "Africa" }, { "bbox": { "b": 164.84762573242188, "coord_origin": "BOTTOMLEFT", "l": 448.7660217285156, "r": 501.6629638671875, "t": 183.8473358154297 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 47, "text": "89,695,835 174,700,245,772" } ] ], "num_cols": 7, "num_rows": 48, "table_cells": [ { "bbox": { "b": 646.2171630859375, "coord_origin": "BOTTOMLEFT", "l": 108.2132339477539, "r": 145.9970703125, "t": 655.6136474609375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 646.3136596679688, "coord_origin": "BOTTOMLEFT", "l": 185.93484497070312, "r": 238.85569763183594, "t": 655.48876953125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "ISO-639-3 catalog-ref Genus" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "" }, { "bbox": { "b": 645.921875, "coord_origin": "BOTTOMLEFT", "l": 353.0313720703125, "r": 381.9716491699219, "t": 655.422607421875 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "Family" }, { "bbox": { "b": 645.62890625, "coord_origin": "BOTTOMLEFT", "l": 406.5981140136719, "r": 444.2469787597656, "t": 655.3299560546875 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "Macroarea" }, { "bbox": { "b": 646.5857543945312, "coord_origin": "BOTTOMLEFT", "l": 463.8620300292969, "r": 500.8646545410156, "t": 656.016845703125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "Size in Bytes" }, { "bbox": { "b": 632.6473388671875, "coord_origin": "BOTTOMLEFT", "l": 107.94133758544922, "r": 127.5891342163086, "t": 641.4132080078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Akan" }, { "bbox": { "b": 632.7122802734375, "coord_origin": "BOTTOMLEFT", "l": 204.35984802246094, "r": 216.82046508789062, "t": 641.340576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "aka" }, { "bbox": { "b": 632.6465454101562, "coord_origin": "BOTTOMLEFT", "l": 244.96432495117188, "r": 254.85003662109375, "t": 640.8958740234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "ak" }, { "bbox": { "b": 632.294677734375, "coord_origin": "BOTTOMLEFT", "l": 272.16265869140625, "r": 290.2823791503906, "t": 640.8880615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "Kwa" }, { "bbox": { "b": 632.7719116210938, "coord_origin": "BOTTOMLEFT", "l": 353.5899963378906, "r": 393.4905700683594, "t": 641.7452392578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "Niger-Congo" }, { "bbox": { "b": 632.2013549804688, "coord_origin": "BOTTOMLEFT", "l": 406.6474609375, "r": 427.2049255371094, "t": 640.8191528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "Africa" }, { "bbox": { "b": 633.165283203125, "coord_origin": "BOTTOMLEFT", "l": 470.9044494628906, "r": 500.62353515625, "t": 641.6571044921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "70,1554" }, { "bbox": { "b": 621.754150390625, "coord_origin": "BOTTOMLEFT", "l": 108.563720703125, "r": 133.888671875, "t": 630.2901000976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Arabic" }, { "bbox": { "b": 622.4773559570312, "coord_origin": "BOTTOMLEFT", "l": 205.65953063964844, "r": 215.71127319335938, "t": 630.92822265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "arb" }, { "bbox": { "b": 622.5323486328125, "coord_origin": "BOTTOMLEFT", "l": 246.74658203125, "r": 253.63404846191406, "t": 630.9849243164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "ar" }, { "bbox": { "b": 622.2947387695312, "coord_origin": "BOTTOMLEFT", "l": 272.299072265625, "r": 291.55072021484375, "t": 630.7213134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "Semitic" }, { "bbox": { "b": 621.9561767578125, "coord_origin": "BOTTOMLEFT", "l": 353.9492492675781, "r": 397.2653503417969, "t": 630.767333984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "Afro-Asiatic" }, { "bbox": { "b": 621.6553344726562, "coord_origin": "BOTTOMLEFT", "l": 406.5619201660156, "r": 428.78033447265625, "t": 630.337158203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 2, "text": "Eurasia" }, { "bbox": { "b": 621.7122802734375, "coord_origin": "BOTTOMLEFT", "l": 454.5549011230469, "r": 500.0721435546875, "t": 629.9695434570312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 2, "text": "74,854,900,600" }, { "bbox": { "b": 611.236328125, "coord_origin": "BOTTOMLEFT", "l": 108.90092468261719, "r": 139.33248901367188, "t": 619.7415771484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Assamese" }, { "bbox": { "b": 611.52978515625, "coord_origin": "BOTTOMLEFT", "l": 204.74105834960938, "r": 216.08767700195312, "t": 619.867431640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "asm" }, { "bbox": { "b": 612.06689453125, "coord_origin": "BOTTOMLEFT", "l": 246.48477172851562, "r": 253.59249877929688, "t": 620.5132446289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "as" }, { "bbox": { "b": 611.5089721679688, "coord_origin": "BOTTOMLEFT", "l": 271.9761962890625, "r": 289.5044860839844, "t": 619.9150390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "Indic" }, { "bbox": { "b": 610.7553100585938, "coord_origin": "BOTTOMLEFT", "l": 353.5668640136719, "r": 391.488037109375, "t": 619.4790649414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 611.9335327148438, "coord_origin": "BOTTOMLEFT", "l": 461.8177795410156, "r": 500.3688049316406, "t": 620.1602172851562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 3, "text": "291,522,098" }, { "bbox": { "b": 601.2576293945312, "coord_origin": "BOTTOMLEFT", "l": 108.98766326904297, "r": 137.57203674316406, "t": 609.6809692382812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Bambara" }, { "bbox": { "b": 601.77294921875, "coord_origin": "BOTTOMLEFT", "l": 203.88540649414062, "r": 216.4199676513672, "t": 610.0787963867188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "bam" }, { "bbox": { "b": 601.263671875, "coord_origin": "BOTTOMLEFT", "l": 245.21353149414062, "r": 254.23228454589844, "t": 609.674560546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "bm" }, { "bbox": { "b": 601.5784301757812, "coord_origin": "BOTTOMLEFT", "l": 274.1773681640625, "r": 318.1032409667969, "t": 610.1167602539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "Western Mande" }, { "bbox": { "b": 600.5735473632812, "coord_origin": "BOTTOMLEFT", "l": 354.537841796875, "r": 378.57391357421875, "t": 609.1177368164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "Mande" }, { "bbox": { "b": 600.7642211914062, "coord_origin": "BOTTOMLEFT", "l": 406.37811279296875, "r": 426.883544921875, "t": 609.1436157226562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 4, "text": "Africa" }, { "bbox": { "b": 601.4264526367188, "coord_origin": "BOTTOMLEFT", "l": 473.25274658203125, "r": 499.7992858886719, "t": 609.6851806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 4, "text": "391,747" }, { "bbox": { "b": 591.3948974609375, "coord_origin": "BOTTOMLEFT", "l": 108.781494140625, "r": 132.4031982421875, "t": 599.66357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 592.23193359375, "coord_origin": "BOTTOMLEFT", "l": 204.74526977539062, "r": 215.89810180664062, "t": 600.4318237304688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "eus" }, { "bbox": { "b": 591.6302490234375, "coord_origin": "BOTTOMLEFT", "l": 245.8119354248047, "r": 254.22149658203125, "t": 599.9959106445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "eu" }, { "bbox": { "b": 591.2930908203125, "coord_origin": "BOTTOMLEFT", "l": 271.3256530761719, "r": 291.7969970703125, "t": 599.5765380859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 590.8136596679688, "coord_origin": "BOTTOMLEFT", "l": 353.91680908203125, "r": 378.7007751464844, "t": 599.1942749023438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 591.1436157226562, "coord_origin": "BOTTOMLEFT", "l": 406.2830505371094, "r": 427.990966796875, "t": 599.5001220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 5, "text": "Eurasia" }, { "bbox": { "b": 591.21923828125, "coord_origin": "BOTTOMLEFT", "l": 458.5888366699219, "r": 500.11090087890625, "t": 599.5518798828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 5, "text": "2,360,470,848" }, { "bbox": { "b": 580.9913330078125, "coord_origin": "BOTTOMLEFT", "l": 108.32389831542969, "r": 132.4408721923828, "t": 589.3953247070312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Bengali" }, { "bbox": { "b": 581.5091552734375, "coord_origin": "BOTTOMLEFT", "l": 204.27264404296875, "r": 216.28842163085938, "t": 589.8223266601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "ben" }, { "bbox": { "b": 581.3764038085938, "coord_origin": "BOTTOMLEFT", "l": 245.8140106201172, "r": 254.11277770996094, "t": 589.6771240234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "bn" }, { "bbox": { "b": 581.5809326171875, "coord_origin": "BOTTOMLEFT", "l": 270.63018798828125, "r": 287.0025634765625, "t": 589.832763671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 6, "text": "Indic" }, { "bbox": { "b": 580.9334106445312, "coord_origin": "BOTTOMLEFT", "l": 354.4245910644531, "r": 399.6710205078125, "t": 589.3593139648438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 6, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 6, "text": "" }, { "bbox": { "b": 580.7463989257812, "coord_origin": "BOTTOMLEFT", "l": 451.1468505859375, "r": 500.87432861328125, "t": 589.0629272460938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 6, "text": "18,606,823,104" }, { "bbox": { "b": 571.077392578125, "coord_origin": "BOTTOMLEFT", "l": 108.18450164794922, "r": 133.05084228515625, "t": 579.47705078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Catalan" }, { "bbox": { "b": 571.3492431640625, "coord_origin": "BOTTOMLEFT", "l": 205.28199768066406, "r": 215.70187377929688, "t": 579.6483154296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "cat" }, { "bbox": { "b": 571.41259765625, "coord_origin": "BOTTOMLEFT", "l": 246.1710968017578, "r": 253.74913024902344, "t": 579.7540283203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "ca" }, { "bbox": { "b": 571.3926391601562, "coord_origin": "BOTTOMLEFT", "l": 272.9360046386719, "r": 303.5674743652344, "t": 579.6792602539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 7, "text": "Romance" }, { "bbox": { "b": 571.1082763671875, "coord_origin": "BOTTOMLEFT", "l": 354.3467102050781, "r": 399.4031677246094, "t": 579.536376953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 7, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 571.4974365234375, "coord_origin": "BOTTOMLEFT", "l": 451.9903869628906, "r": 500.81103515625, "t": 579.8428955078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 7, "text": "17,792,493,289" }, { "bbox": { "b": 560.6057739257812, "coord_origin": "BOTTOMLEFT", "l": 108.31121826171875, "r": 143.47100830078125, "t": 569.05322265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Chi Chewa" }, { "bbox": { "b": 560.7371826171875, "coord_origin": "BOTTOMLEFT", "l": 204.81883239746094, "r": 216.561279296875, "t": 569.0138549804688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "nya" }, { "bbox": { "b": 560.5833740234375, "coord_origin": "BOTTOMLEFT", "l": 245.78848266601562, "r": 253.6044921875, "t": 568.9688720703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "ny" }, { "bbox": { "b": 560.8050537109375, "coord_origin": "BOTTOMLEFT", "l": 271.7248840332031, "r": 298.0384216308594, "t": 568.9901733398438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 8, "text": "Bantoid" }, { "bbox": { "b": 560.1963500976562, "coord_origin": "BOTTOMLEFT", "l": 354.5111389160156, "r": 394.291015625, "t": 568.4911499023438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 8, "text": "Niger-Congo" }, { "bbox": { "b": 560.3571166992188, "coord_origin": "BOTTOMLEFT", "l": 406.7816467285156, "r": 427.1576232910156, "t": 568.5674438476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 8, "text": "Africa" }, { "bbox": { "b": 560.4910278320312, "coord_origin": "BOTTOMLEFT", "l": 468.42236328125, "r": 501.07012939453125, "t": 568.7451782226562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 8, "text": "1,187,405" }, { "bbox": { "b": 550.3297729492188, "coord_origin": "BOTTOMLEFT", "l": 108.46611785888672, "r": 142.48959350585938, "t": 558.7322998046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Chi Shona" }, { "bbox": { "b": 550.3215942382812, "coord_origin": "BOTTOMLEFT", "l": 205.2152862548828, "r": 216.55426025390625, "t": 558.6187133789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "sna" }, { "bbox": { "b": 550.130859375, "coord_origin": "BOTTOMLEFT", "l": 246.00352478027344, "r": 253.7533416748047, "t": 558.5576782226562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "sn" }, { "bbox": { "b": 550.825927734375, "coord_origin": "BOTTOMLEFT", "l": 271.8756103515625, "r": 297.1409606933594, "t": 559.0391235351562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 9, "text": "Bantoid" }, { "bbox": { "b": 550.585205078125, "coord_origin": "BOTTOMLEFT", "l": 354.7015075683594, "r": 393.3777160644531, "t": 558.9050903320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 9, "text": "Niger-Congo" }, { "bbox": { "b": 550.6567993164062, "coord_origin": "BOTTOMLEFT", "l": 406.7747802734375, "r": 427.2490539550781, "t": 558.8569946289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 9, "text": "Africa" }, { "bbox": { "b": 550.6953125, "coord_origin": "BOTTOMLEFT", "l": 467.956787109375, "r": 500.91748046875, "t": 558.9978637695312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 9, "text": "6,638,639" }, { "bbox": { "b": 540.658447265625, "coord_origin": "BOTTOMLEFT", "l": 108.03372955322266, "r": 152.3150177001953, "t": 549.0789794921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Chi Tumbuka" }, { "bbox": { "b": 540.6922607421875, "coord_origin": "BOTTOMLEFT", "l": 204.47483825683594, "r": 216.28294372558594, "t": 548.9986572265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "tum" }, { "bbox": { "b": 540.871337890625, "coord_origin": "BOTTOMLEFT", "l": 244.4879608154297, "r": 254.66249084472656, "t": 549.2894287109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "tum" }, { "bbox": { "b": 541.38818359375, "coord_origin": "BOTTOMLEFT", "l": 270.99932861328125, "r": 295.212646484375, "t": 549.5797729492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 10, "text": "Bantoid" }, { "bbox": { "b": 540.8161010742188, "coord_origin": "BOTTOMLEFT", "l": 354.9844970703125, "r": 395.4344787597656, "t": 549.1530151367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 10, "text": "Niger-Congo" }, { "bbox": { "b": 540.8040771484375, "coord_origin": "BOTTOMLEFT", "l": 406.79461669921875, "r": 427.968994140625, "t": 549.0009765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 10, "text": "Africa" }, { "bbox": { "b": 540.8871459960938, "coord_origin": "BOTTOMLEFT", "l": 471.609130859375, "r": 501.21917724609375, "t": 549.2110595703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 10, "text": "170,360" }, { "bbox": { "b": 530.3832397460938, "coord_origin": "BOTTOMLEFT", "l": 108.039794921875, "r": 129.19537353515625, "t": 538.6337280273438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "English" }, { "bbox": { "b": 530.1448364257812, "coord_origin": "BOTTOMLEFT", "l": 204.3148193359375, "r": 216.1548614501953, "t": 538.4827880859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "eng" }, { "bbox": { "b": 529.9124145507812, "coord_origin": "BOTTOMLEFT", "l": 244.63050842285156, "r": 254.49581909179688, "t": 538.3505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "en" }, { "bbox": { "b": 530.45556640625, "coord_origin": "BOTTOMLEFT", "l": 272.2205810546875, "r": 301.80712890625, "t": 538.64990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 11, "text": "Germanic" }, { "bbox": { "b": 530.1826171875, "coord_origin": "BOTTOMLEFT", "l": 354.6688537597656, "r": 398.07647705078125, "t": 538.549560546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 11, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 530.55322265625, "coord_origin": "BOTTOMLEFT", "l": 457.59869384765625, "r": 500.9771423339844, "t": 538.8646850585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 11, "text": "484,953,009,124" }, { "bbox": { "b": 520.1136474609375, "coord_origin": "BOTTOMLEFT", "l": 107.9852294921875, "r": 124.80977630615234, "t": 528.241943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Fon" }, { "bbox": { "b": 520.451416015625, "coord_origin": "BOTTOMLEFT", "l": 205.04534912109375, "r": 215.4799346923828, "t": 528.7557983398438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "fon" }, { "bbox": { "b": 520.3941650390625, "coord_origin": "BOTTOMLEFT", "l": 245.62103271484375, "r": 254.02305603027344, "t": 528.8161010742188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "fon" }, { "bbox": { "b": 520.4472045898438, "coord_origin": "BOTTOMLEFT", "l": 272.12701416015625, "r": 287.24969482421875, "t": 528.5364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 12, "text": "Kwa" }, { "bbox": { "b": 520.2186279296875, "coord_origin": "BOTTOMLEFT", "l": 354.25396728515625, "r": 396.1515808105469, "t": 528.5382080078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 12, "text": "Niger-Congo" }, { "bbox": { "b": 520.4370727539062, "coord_origin": "BOTTOMLEFT", "l": 406.46966552734375, "r": 428.5343933105469, "t": 528.7864379882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 12, "text": "Africa" }, { "bbox": { "b": 520.3253173828125, "coord_origin": "BOTTOMLEFT", "l": 464.21484375, "r": 501.8062438964844, "t": 528.512451171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 12, "text": "2,478,546" }, { "bbox": { "b": 510.3376770019531, "coord_origin": "BOTTOMLEFT", "l": 108.0778579711914, "r": 128.46267700195312, "t": 518.528076171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "French" }, { "bbox": { "b": 510.6510314941406, "coord_origin": "BOTTOMLEFT", "l": 204.95750427246094, "r": 215.9420166015625, "t": 519.1000366210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "fra" }, { "bbox": { "b": 510.6396484375, "coord_origin": "BOTTOMLEFT", "l": 246.32420349121094, "r": 253.78955078125, "t": 519.1185302734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "fr" }, { "bbox": { "b": 510.6376953125, "coord_origin": "BOTTOMLEFT", "l": 272.39910888671875, "r": 300.5570373535156, "t": 518.8150634765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 13, "text": "Romance" }, { "bbox": { "b": 510.453369140625, "coord_origin": "BOTTOMLEFT", "l": 354.68988037109375, "r": 400.3841552734375, "t": 518.8603515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 13, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 13, "text": "" }, { "bbox": { "b": 500.30072021484375, "coord_origin": "BOTTOMLEFT", "l": 109.16815185546875, "r": 134.55955505371094, "t": 508.5826721191406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "Gujarati" }, { "bbox": { "b": 500.5307312011719, "coord_origin": "BOTTOMLEFT", "l": 205.54954528808594, "r": 215.7245330810547, "t": 508.9515075683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "guj" }, { "bbox": { "b": 500.3470153808594, "coord_origin": "BOTTOMLEFT", "l": 246.97833251953125, "r": 253.26995849609375, "t": 508.82855224609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "gu" }, { "bbox": { "b": 500.1457824707031, "coord_origin": "BOTTOMLEFT", "l": 272.1981201171875, "r": 288.41845703125, "t": 508.32208251953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 14, "text": "Indic" }, { "bbox": { "b": 499.9831848144531, "coord_origin": "BOTTOMLEFT", "l": 354.54937744140625, "r": 401.7975158691406, "t": 508.4107666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 14, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 499.7450256347656, "coord_origin": "BOTTOMLEFT", "l": 452.34222412109375, "r": 501.66302490234375, "t": 516.962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 14, "text": "208,242,620,434 1,199,986,460" }, { "bbox": { "b": 490.5615234375, "coord_origin": "BOTTOMLEFT", "l": 108.0564956665039, "r": 128.54466247558594, "t": 498.7920227050781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "Hindi" }, { "bbox": { "b": 490.3254699707031, "coord_origin": "BOTTOMLEFT", "l": 205.20111083984375, "r": 216.18690490722656, "t": 498.85760498046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "hin" }, { "bbox": { "b": 490.57574462890625, "coord_origin": "BOTTOMLEFT", "l": 246.58816528320312, "r": 253.3518524169922, "t": 499.0936279296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "hi" }, { "bbox": { "b": 490.6678771972656, "coord_origin": "BOTTOMLEFT", "l": 271.8505554199219, "r": 289.2873840332031, "t": 498.8848571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 15, "text": "Indic" }, { "bbox": { "b": 490.0447082519531, "coord_origin": "BOTTOMLEFT", "l": 354.6220703125, "r": 400.40203857421875, "t": 498.503662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 15, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 15, "text": "" }, { "bbox": { "b": 490.8396301269531, "coord_origin": "BOTTOMLEFT", "l": 454.7178955078125, "r": 501.1356201171875, "t": 499.1708984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 15, "text": "24,622,119,985" }, { "bbox": { "b": 480.4582824707031, "coord_origin": "BOTTOMLEFT", "l": 108.25355529785156, "r": 123.8819808959961, "t": 488.71112060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "Igbo" }, { "bbox": { "b": 480.2000427246094, "coord_origin": "BOTTOMLEFT", "l": 205.50733947753906, "r": 216.15029907226562, "t": 488.73748779296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "ibo" }, { "bbox": { "b": 480.3326416015625, "coord_origin": "BOTTOMLEFT", "l": 246.79440307617188, "r": 253.29974365234375, "t": 488.8615417480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "ig" }, { "bbox": { "b": 480.50146484375, "coord_origin": "BOTTOMLEFT", "l": 271.8854675292969, "r": 291.75390625, "t": 488.7595520019531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 16, "text": "Igboid" }, { "bbox": { "b": 479.901611328125, "coord_origin": "BOTTOMLEFT", "l": 354.8074645996094, "r": 396.034912109375, "t": 488.3544006347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 16, "text": "Niger-Congo" }, { "bbox": { "b": 480.2947998046875, "coord_origin": "BOTTOMLEFT", "l": 406.9546813964844, "r": 427.10858154296875, "t": 488.5445251464844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 16, "text": "Africa" }, { "bbox": { "b": 480.22882080078125, "coord_origin": "BOTTOMLEFT", "l": 467.469482421875, "r": 502.2340393066406, "t": 488.60491943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 16, "text": "14078,521" }, { "bbox": { "b": 470.4344177246094, "coord_origin": "BOTTOMLEFT", "l": 108.68071746826172, "r": 137.19676208496094, "t": 478.73614501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "Indonesian" }, { "bbox": { "b": 470.5469970703125, "coord_origin": "BOTTOMLEFT", "l": 204.8577423095703, "r": 216.13925170898438, "t": 479.11810302734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "ind" }, { "bbox": { "b": 470.5489501953125, "coord_origin": "BOTTOMLEFT", "l": 246.34152221679688, "r": 253.47520446777344, "t": 479.0879821777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "id" }, { "bbox": { "b": 470.4266357421875, "coord_origin": "BOTTOMLEFT", "l": 273.7782897949219, "r": 325.0870056152344, "t": 478.8460693359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 17, "text": "Malayo-Sumbawan" }, { "bbox": { "b": 470.0202941894531, "coord_origin": "BOTTOMLEFT", "l": 355.7085266113281, "r": 395.0197448730469, "t": 478.4385681152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 17, "text": "Austronesian" }, { "bbox": { "b": 470.5386047363281, "coord_origin": "BOTTOMLEFT", "l": 406.584716796875, "r": 430.518798828125, "t": 478.8246154785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 17, "text": "Papunesia" }, { "bbox": { "b": 470.53857421875, "coord_origin": "BOTTOMLEFT", "l": 457.6726379394531, "r": 501.431884765625, "t": 478.9574279785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 17, "text": "19,972,325,222" }, { "bbox": { "b": 460.09820556640625, "coord_origin": "BOTTOMLEFT", "l": 108.10408020019531, "r": 132.24937438964844, "t": 468.4632568359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "Isi Zulu" }, { "bbox": { "b": 460.4718933105469, "coord_origin": "BOTTOMLEFT", "l": 204.99404907226562, "r": 216.0699005126953, "t": 469.08599853515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "zul" }, { "bbox": { "b": 460.7010498046875, "coord_origin": "BOTTOMLEFT", "l": 246.2052001953125, "r": 253.5150909423828, "t": 469.2560729980469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "zu" }, { "bbox": { "b": 461.1459045410156, "coord_origin": "BOTTOMLEFT", "l": 271.419677734375, "r": 299.9014587402344, "t": 469.4657897949219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 18, "text": "Bantoid" }, { "bbox": { "b": 460.0359191894531, "coord_origin": "BOTTOMLEFT", "l": 355.341064453125, "r": 392.1656494140625, "t": 468.4740295410156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 18, "text": "Niger-Congo" }, { "bbox": { "b": 460.8236389160156, "coord_origin": "BOTTOMLEFT", "l": 407.4127502441406, "r": 427.576171875, "t": 469.1214904785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 18, "text": "Africa" }, { "bbox": { "b": 460.66864013671875, "coord_origin": "BOTTOMLEFT", "l": 468.42584228515625, "r": 501.05413818359375, "t": 469.1996765136719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 18, "text": "8,511,561" }, { "bbox": { "b": 450.20770263671875, "coord_origin": "BOTTOMLEFT", "l": 108.04843139648438, "r": 136.48728942871094, "t": 458.585205078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "Kannada" }, { "bbox": { "b": 449.7711181640625, "coord_origin": "BOTTOMLEFT", "l": 204.49652099609375, "r": 216.16058349609375, "t": 458.3982238769531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "kan" }, { "bbox": { "b": 449.9278869628906, "coord_origin": "BOTTOMLEFT", "l": 244.67210388183594, "r": 253.61407470703125, "t": 458.5023498535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "kn" }, { "bbox": { "b": 450.34820556640625, "coord_origin": "BOTTOMLEFT", "l": 272.7766418457031, "r": 336.11004638671875, "t": 458.8428039550781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 19, "text": "Southern Dravidian" }, { "bbox": { "b": 449.8624267578125, "coord_origin": "BOTTOMLEFT", "l": 355.93853759765625, "r": 395.9790344238281, "t": 458.3201904296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 19, "text": "Dravidian" }, { "bbox": { "b": 450.1551513671875, "coord_origin": "BOTTOMLEFT", "l": 406.7572937011719, "r": 429.6174621582031, "t": 458.4588623046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 19, "text": "Eurasia" }, { "bbox": { "b": 450.2169494628906, "coord_origin": "BOTTOMLEFT", "l": 457.0106201171875, "r": 501.13494873046875, "t": 458.651123046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 19, "text": "2,098,453,560" }, { "bbox": { "b": 439.8309631347656, "coord_origin": "BOTTOMLEFT", "l": 108.59833526611328, "r": 134.7786865234375, "t": 448.2895812988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "Kikuyu" }, { "bbox": { "b": 439.64404296875, "coord_origin": "BOTTOMLEFT", "l": 204.66603088378906, "r": 216.0258026123047, "t": 448.2994079589844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "kik" }, { "bbox": { "b": 440.22259521484375, "coord_origin": "BOTTOMLEFT", "l": 245.57017517089844, "r": 253.76902770996094, "t": 448.8172912597656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 20, "text": "ki" }, { "bbox": { "b": 440.0103759765625, "coord_origin": "BOTTOMLEFT", "l": 271.54241943359375, "r": 297.26104736328125, "t": 448.4076843261719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 20, "text": "Bantoid" }, { "bbox": { "b": 439.05694580078125, "coord_origin": "BOTTOMLEFT", "l": 354.9132995605469, "r": 394.5080871582031, "t": 447.5910949707031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 20, "text": "Niger-Congo" }, { "bbox": { "b": 440.0498962402344, "coord_origin": "BOTTOMLEFT", "l": 407.20941162109375, "r": 428.545166015625, "t": 448.4403076171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 20, "text": "Africa" }, { "bbox": { "b": 439.6396789550781, "coord_origin": "BOTTOMLEFT", "l": 466.3895568847656, "r": 500.88665771484375, "t": 448.1282043457031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 20, "text": "359,615" }, { "bbox": { "b": 429.197509765625, "coord_origin": "BOTTOMLEFT", "l": 108.9115982055664, "r": 154.6288604736328, "t": 437.7027587890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 21, "text": "Kinyarwanda" }, { "bbox": { "b": 429.2398681640625, "coord_origin": "BOTTOMLEFT", "l": 204.80714416503906, "r": 215.96336364746094, "t": 437.84503173828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 21, "text": "kin" }, { "bbox": { "b": 430.0733947753906, "coord_origin": "BOTTOMLEFT", "l": 245.4805145263672, "r": 253.70411682128906, "t": 438.66058349609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 21, "text": "rw" }, { "bbox": { "b": 429.03387451171875, "coord_origin": "BOTTOMLEFT", "l": 271.88140869140625, "r": 296.1385803222656, "t": 437.34600830078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 21, "text": "Bantoid" }, { "bbox": { "b": 429.1513977050781, "coord_origin": "BOTTOMLEFT", "l": 354.9581604003906, "r": 393.7857360839844, "t": 437.66802978515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 21, "text": "Niger-Congo" }, { "bbox": { "b": 429.6646728515625, "coord_origin": "BOTTOMLEFT", "l": 407.10028076171875, "r": 428.4424133300781, "t": 438.06573486328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 21, "text": "Africa" }, { "bbox": { "b": 429.5914001464844, "coord_origin": "BOTTOMLEFT", "l": 466.16143798828125, "r": 500.904052734375, "t": 438.051025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 22, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 21, "text": "40,428,299" }, { "bbox": { "b": 418.99176025390625, "coord_origin": "BOTTOMLEFT", "l": 109.14940643310547, "r": 135.82835388183594, "t": 427.58953857421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 22, "text": "Kirundi" }, { "bbox": { "b": 419.1227111816406, "coord_origin": "BOTTOMLEFT", "l": 204.61392211914062, "r": 216.1142578125, "t": 427.8158264160156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 22, "text": "run" }, { "bbox": { "b": 419.64666748046875, "coord_origin": "BOTTOMLEFT", "l": 245.8270263671875, "r": 253.95115661621094, "t": 428.2518005371094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 22, "text": "rn" }, { "bbox": { "b": 419.5191345214844, "coord_origin": "BOTTOMLEFT", "l": 272.36932373046875, "r": 297.0074157714844, "t": 427.9472961425781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 22, "text": "Bantoid" }, { "bbox": { "b": 418.98175048828125, "coord_origin": "BOTTOMLEFT", "l": 354.773681640625, "r": 394.40399169921875, "t": 427.59405517578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 22, "text": "Niger-Congo" }, { "bbox": { "b": 419.12158203125, "coord_origin": "BOTTOMLEFT", "l": 407.1785888671875, "r": 428.7720947265625, "t": 427.5682067871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 22, "text": "Africa" }, { "bbox": { "b": 419.4420166015625, "coord_origin": "BOTTOMLEFT", "l": 467.3368835449219, "r": 500.8492736816406, "t": 427.9061584472656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 23, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 22, "text": "3,272,550" }, { "bbox": { "b": 409.0594787597656, "coord_origin": "BOTTOMLEFT", "l": 108.30078887939453, "r": 131.46534729003906, "t": 417.6105651855469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 23, "text": "Lingala" }, { "bbox": { "b": 408.3804016113281, "coord_origin": "BOTTOMLEFT", "l": 205.62677001953125, "r": 215.6280517578125, "t": 416.9595947265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 23, "text": "lin" }, { "bbox": { "b": 409.2570495605469, "coord_origin": "BOTTOMLEFT", "l": 246.25428771972656, "r": 253.78628540039062, "t": 417.85589599609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 23, "text": "ln" }, { "bbox": { "b": 408.9028625488281, "coord_origin": "BOTTOMLEFT", "l": 271.39996337890625, "r": 294.73968505859375, "t": 417.22637939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 23, "text": "Bantoid" }, { "bbox": { "b": 409.0890808105469, "coord_origin": "BOTTOMLEFT", "l": 355.3092346191406, "r": 393.3785705566406, "t": 417.5863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 23, "text": "Niger-Congo" }, { "bbox": { "b": 408.7477722167969, "coord_origin": "BOTTOMLEFT", "l": 407.39764404296875, "r": 427.63739013671875, "t": 417.1451110839844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 23, "text": "Africa" }, { "bbox": { "b": 409.3039245605469, "coord_origin": "BOTTOMLEFT", "l": 468.5851745605469, "r": 500.78265380859375, "t": 417.762451171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 24, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 23, "text": "1,650,804" }, { "bbox": { "b": 398.44500732421875, "coord_origin": "BOTTOMLEFT", "l": 108.74641418457031, "r": 137.70664978027344, "t": 407.0060119628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 24, "text": "Luganda" }, { "bbox": { "b": 398.3811950683594, "coord_origin": "BOTTOMLEFT", "l": 204.6643524169922, "r": 215.9978790283203, "t": 407.0416564941406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 24, "text": "lug" }, { "bbox": { "b": 399.0399475097656, "coord_origin": "BOTTOMLEFT", "l": 246.8289794921875, "r": 254.31515502929688, "t": 407.6708984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 24, "text": "lg" }, { "bbox": { "b": 398.84063720703125, "coord_origin": "BOTTOMLEFT", "l": 271.9038391113281, "r": 296.4460754394531, "t": 407.2830810546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 24, "text": "Bantoid" }, { "bbox": { "b": 398.667236328125, "coord_origin": "BOTTOMLEFT", "l": 354.76171875, "r": 394.0806579589844, "t": 407.2316589355469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 24, "text": "Niger-Congo" }, { "bbox": { "b": 398.8894348144531, "coord_origin": "BOTTOMLEFT", "l": 407.5158996582031, "r": 427.80133056640625, "t": 407.3270263671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 24, "text": "Africa" }, { "bbox": { "b": 399.0642395019531, "coord_origin": "BOTTOMLEFT", "l": 466.9955749511719, "r": 501.0726623535156, "t": 407.54583740234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 25, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 24, "text": "4,568,367" }, { "bbox": { "b": 388.86865234375, "coord_origin": "BOTTOMLEFT", "l": 108.54914855957031, "r": 141.93531799316406, "t": 397.4731140136719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 25, "text": "Malayalam" }, { "bbox": { "b": 388.5834045410156, "coord_origin": "BOTTOMLEFT", "l": 204.49880981445312, "r": 216.2100067138672, "t": 397.2698059082031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 25, "text": "mal" }, { "bbox": { "b": 388.9284973144531, "coord_origin": "BOTTOMLEFT", "l": 245.669677734375, "r": 253.9735870361328, "t": 397.54095458984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 25, "text": "ml" }, { "bbox": { "b": 388.9745178222656, "coord_origin": "BOTTOMLEFT", "l": 271.04913330078125, "r": 310.7453918457031, "t": 397.46856689453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 25, "text": "Southern Dravidian" }, { "bbox": { "b": 388.6893310546875, "coord_origin": "BOTTOMLEFT", "l": 355.494140625, "r": 394.22955322265625, "t": 397.2818908691406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 25, "text": "Dravidian" }, { "bbox": { "b": 388.61322021484375, "coord_origin": "BOTTOMLEFT", "l": 407.1199035644531, "r": 429.5668029785156, "t": 397.1412353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 25, "text": "Eurasia" }, { "bbox": { "b": 388.4208679199219, "coord_origin": "BOTTOMLEFT", "l": 458.1934509277344, "r": 501.4151306152344, "t": 396.88128662109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 26, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 25, "text": "3,662,571,498" }, { "bbox": { "b": 378.7137756347656, "coord_origin": "BOTTOMLEFT", "l": 109.17071533203125, "r": 135.9722442626953, "t": 387.33428955078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 26, "text": "Marathi" }, { "bbox": { "b": 377.8516845703125, "coord_origin": "BOTTOMLEFT", "l": 204.9829864501953, "r": 216.25836181640625, "t": 386.425048828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 26, "text": "mar" }, { "bbox": { "b": 378.8268737792969, "coord_origin": "BOTTOMLEFT", "l": 245.8630828857422, "r": 253.68673706054688, "t": 387.4132385253906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 26, "text": "mr" }, { "bbox": { "b": 378.7937927246094, "coord_origin": "BOTTOMLEFT", "l": 270.7461242675781, "r": 292.65325927734375, "t": 387.2211608886719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 26, "text": "Indic" }, { "bbox": { "b": 378.4562683105469, "coord_origin": "BOTTOMLEFT", "l": 355.687255859375, "r": 396.0417785644531, "t": 387.03192138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 26, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 26, "text": "" }, { "bbox": { "b": 377.6427917480469, "coord_origin": "BOTTOMLEFT", "l": 454.1509094238281, "r": 501.5225830078125, "t": 386.1326904296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 27, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 26, "text": "1,775,483,122" }, { "bbox": { "b": 368.8480529785156, "coord_origin": "BOTTOMLEFT", "l": 108.32581329345703, "r": 128.98614501953125, "t": 377.3013916015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 27, "text": "Nepali" }, { "bbox": { "b": 368.4093322753906, "coord_origin": "BOTTOMLEFT", "l": 205.1424102783203, "r": 216.29241943359375, "t": 377.0423889160156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 27, "text": "nep" }, { "bbox": { "b": 368.9896240234375, "coord_origin": "BOTTOMLEFT", "l": 245.89109802246094, "r": 253.8175811767578, "t": 377.5635681152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 27, "text": "ne" }, { "bbox": { "b": 369.712890625, "coord_origin": "BOTTOMLEFT", "l": 270.24749755859375, "r": 293.37432861328125, "t": 378.1374816894531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 27, "text": "Indic" }, { "bbox": { "b": 368.70111083984375, "coord_origin": "BOTTOMLEFT", "l": 355.7784118652344, "r": 398.30487060546875, "t": 377.30120849609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 27, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 27, "text": "" }, { "bbox": { "b": 368.2612609863281, "coord_origin": "BOTTOMLEFT", "l": 455.61175537109375, "r": 501.53887939453125, "t": 376.69549560546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 28, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 27, "text": "2,551,307,393" }, { "bbox": { "b": 357.67095947265625, "coord_origin": "BOTTOMLEFT", "l": 109.35924530029297, "r": 155.69496154785156, "t": 366.2355041503906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 28, "text": "Northern Sotho" }, { "bbox": { "b": 358.3829650878906, "coord_origin": "BOTTOMLEFT", "l": 205.17263793945312, "r": 216.2655029296875, "t": 366.9328308105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 28, "text": "nso" }, { "bbox": { "b": 358.58172607421875, "coord_origin": "BOTTOMLEFT", "l": 245.02195739746094, "r": 253.70367431640625, "t": 367.1554260253906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 28, "text": "nso" }, { "bbox": { "b": 358.9200134277344, "coord_origin": "BOTTOMLEFT", "l": 271.0533142089844, "r": 293.208740234375, "t": 367.3212585449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 28, "text": "Bantoid" }, { "bbox": { "b": 358.2391052246094, "coord_origin": "BOTTOMLEFT", "l": 355.3683776855469, "r": 396.8522644042969, "t": 366.7705993652344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 28, "text": "Niger-Congo" }, { "bbox": { "b": 358.4185791015625, "coord_origin": "BOTTOMLEFT", "l": 407.396728515625, "r": 428.9354553222656, "t": 366.8617858886719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 28, "text": "Africa" }, { "bbox": { "b": 358.656005859375, "coord_origin": "BOTTOMLEFT", "l": 467.3802490234375, "r": 500.82958984375, "t": 367.0387878417969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 29, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 28, "text": "1,764,506" }, { "bbox": { "b": 348.6308898925781, "coord_origin": "BOTTOMLEFT", "l": 108.28306579589844, "r": 126.74258422851562, "t": 357.0397644042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 29, "text": "Odia" }, { "bbox": { "b": 348.3537902832031, "coord_origin": "BOTTOMLEFT", "l": 204.96728515625, "r": 216.32981872558594, "t": 356.9362487792969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 29, "text": "ori" }, { "bbox": { "b": 349.1275939941406, "coord_origin": "BOTTOMLEFT", "l": 245.6078338623047, "r": 253.9015655517578, "t": 357.6670227050781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 29, "text": "or" }, { "bbox": { "b": 349.31854248046875, "coord_origin": "BOTTOMLEFT", "l": 270.2909240722656, "r": 291.72900390625, "t": 357.7449035644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 29, "text": "Indic" }, { "bbox": { "b": 348.29119873046875, "coord_origin": "BOTTOMLEFT", "l": 355.2393798828125, "r": 400.64923095703125, "t": 356.9054870605469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 29, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 29, "text": "" }, { "bbox": { "b": 348.8809509277344, "coord_origin": "BOTTOMLEFT", "l": 455.0662841796875, "r": 501.3878173828125, "t": 357.31317138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 30, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 29, "text": "1,157,100,133" }, { "bbox": { "b": 338.6387634277344, "coord_origin": "BOTTOMLEFT", "l": 108.17900848388672, "r": 140.20892333984375, "t": 347.0970458984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 30, "text": "Portuguese" }, { "bbox": { "b": 338.3876953125, "coord_origin": "BOTTOMLEFT", "l": 205.31768798828125, "r": 216.0242156982422, "t": 346.7798767089844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 30, "text": "por" }, { "bbox": { "b": 338.287353515625, "coord_origin": "BOTTOMLEFT", "l": 246.67745971679688, "r": 253.61895751953125, "t": 346.75762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 30, "text": "pt" }, { "bbox": { "b": 339.12713623046875, "coord_origin": "BOTTOMLEFT", "l": 270.8980407714844, "r": 298.5016174316406, "t": 347.4864501953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 30, "text": "Romance" }, { "bbox": { "b": 337.8818664550781, "coord_origin": "BOTTOMLEFT", "l": 355.2070007324219, "r": 401.1482238769531, "t": 346.43310546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 30, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 30, "text": "" }, { "bbox": { "b": 338.24639892578125, "coord_origin": "BOTTOMLEFT", "l": 453.4613952636719, "r": 501.1967468261719, "t": 346.6802062988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 31, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 30, "text": "79,277,543,375" }, { "bbox": { "b": 327.7706604003906, "coord_origin": "BOTTOMLEFT", "l": 108.52886962890625, "r": 132.45672607421875, "t": 336.1958312988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 31, "text": "Punjabi" }, { "bbox": { "b": 328.3049621582031, "coord_origin": "BOTTOMLEFT", "l": 205.0506134033203, "r": 216.06906127929688, "t": 336.6721496582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 31, "text": "pan" }, { "bbox": { "b": 328.44378662109375, "coord_origin": "BOTTOMLEFT", "l": 246.59326171875, "r": 253.76263427734375, "t": 336.8516845703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 31, "text": "pa" }, { "bbox": { "b": 328.44622802734375, "coord_origin": "BOTTOMLEFT", "l": 271.5168762207031, "r": 286.8114929199219, "t": 336.776611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 31, "text": "Indic" }, { "bbox": { "b": 327.73773193359375, "coord_origin": "BOTTOMLEFT", "l": 355.3795166015625, "r": 399.90777587890625, "t": 336.2171325683594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 31, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 31, "text": "" }, { "bbox": { "b": 328.4530334472656, "coord_origin": "BOTTOMLEFT", "l": 462.5596008300781, "r": 500.4649963378906, "t": 336.83160400390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 32, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 31, "text": "1,572,109,752" }, { "bbox": { "b": 317.9649658203125, "coord_origin": "BOTTOMLEFT", "l": 108.49153900146484, "r": 134.1531524658203, "t": 326.4454345703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 32, "text": "Sesotho" }, { "bbox": { "b": 317.9165344238281, "coord_origin": "BOTTOMLEFT", "l": 205.4767608642578, "r": 215.55690002441406, "t": 326.3720397949219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 32, "text": "sot" }, { "bbox": { "b": 318.0326843261719, "coord_origin": "BOTTOMLEFT", "l": 246.9828338623047, "r": 253.73391723632812, "t": 326.5728759765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 32, "text": "st" }, { "bbox": { "b": 318.3270263671875, "coord_origin": "BOTTOMLEFT", "l": 272.4049072265625, "r": 295.37548828125, "t": 326.66546630859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 32, "text": "Bantoid" }, { "bbox": { "b": 317.82415771484375, "coord_origin": "BOTTOMLEFT", "l": 354.586181640625, "r": 396.1689758300781, "t": 326.36676025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 32, "text": "Niger-Congo" }, { "bbox": { "b": 318.34857177734375, "coord_origin": "BOTTOMLEFT", "l": 407.21234130859375, "r": 427.7130432128906, "t": 326.7799072265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 32, "text": "Africa" }, { "bbox": { "b": 318.2388916015625, "coord_origin": "BOTTOMLEFT", "l": 473.3670654296875, "r": 500.52459716796875, "t": 326.6830749511719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 33, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 32, "text": "751,034" }, { "bbox": { "b": 307.5071105957031, "coord_origin": "BOTTOMLEFT", "l": 108.30636596679688, "r": 136.1265411376953, "t": 315.8783264160156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 33, "text": "Setswana" }, { "bbox": { "b": 307.56317138671875, "coord_origin": "BOTTOMLEFT", "l": 205.66221618652344, "r": 215.0456085205078, "t": 316.0390930175781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 33, "text": "tsn" }, { "bbox": { "b": 308.10260009765625, "coord_origin": "BOTTOMLEFT", "l": 247.04190063476562, "r": 253.63511657714844, "t": 316.6589660644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 33, "text": "tn" }, { "bbox": { "b": 307.83221435546875, "coord_origin": "BOTTOMLEFT", "l": 273.2046813964844, "r": 296.1052551269531, "t": 316.2181701660156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 33, "text": "Bantoid" }, { "bbox": { "b": 307.2530822753906, "coord_origin": "BOTTOMLEFT", "l": 353.7736511230469, "r": 392.70672607421875, "t": 315.7368469238281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 33, "text": "Niger-Congo" }, { "bbox": { "b": 307.73291015625, "coord_origin": "BOTTOMLEFT", "l": 407.1366271972656, "r": 427.6194152832031, "t": 316.18487548828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 33, "text": "Africa" }, { "bbox": { "b": 308.2646484375, "coord_origin": "BOTTOMLEFT", "l": 471.7481384277344, "r": 500.15313720703125, "t": 316.6508483886719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 34, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 33, "text": "1,502,200" }, { "bbox": { "b": 297.7591247558594, "coord_origin": "BOTTOMLEFT", "l": 108.14258575439453, "r": 169.5782928466797, "t": 306.10028076171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 34, "text": "Simplified Chinese" }, { "bbox": { "b": 297.88739013671875, "coord_origin": "BOTTOMLEFT", "l": 205.19297790527344, "r": 215.63514709472656, "t": 306.369384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 34, "text": "-" }, { "bbox": { "b": 298.06085205078125, "coord_origin": "BOTTOMLEFT", "l": 244.327392578125, "r": 254.46121215820312, "t": 306.5933532714844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 34, "text": "zhs" }, { "bbox": { "b": 298.551513671875, "coord_origin": "BOTTOMLEFT", "l": 271.9973449707031, "r": 298.54327392578125, "t": 306.8904724121094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 34, "text": "Chinese" }, { "bbox": { "b": 297.8870849609375, "coord_origin": "BOTTOMLEFT", "l": 354.0538024902344, "r": 396.21484375, "t": 306.44537353515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 34, "text": "Sino-Tibetan" }, { "bbox": { "b": 297.8769226074219, "coord_origin": "BOTTOMLEFT", "l": 406.57568359375, "r": 429.7794189453125, "t": 306.42022705078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 34, "text": "Eurasia" }, { "bbox": { "b": 297.3220520019531, "coord_origin": "BOTTOMLEFT", "l": 450.81640625, "r": 500.7181396484375, "t": 305.8003234863281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 35, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 34, "text": "261,019,433,892" }, { "bbox": { "b": 287.2420654296875, "coord_origin": "BOTTOMLEFT", "l": 108.35747528076172, "r": 134.9157257080078, "t": 295.62994384765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 35, "text": "Spanish" }, { "bbox": { "b": 287.7139587402344, "coord_origin": "BOTTOMLEFT", "l": 205.1435089111328, "r": 215.99790954589844, "t": 296.11444091796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 35, "text": "spa" }, { "bbox": { "b": 287.77813720703125, "coord_origin": "BOTTOMLEFT", "l": 245.9720458984375, "r": 254.34469604492188, "t": 296.24041748046875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 35, "text": "es" }, { "bbox": { "b": 287.5814208984375, "coord_origin": "BOTTOMLEFT", "l": 271.341796875, "r": 301.1479797363281, "t": 295.8623352050781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 35, "text": "Romance" }, { "bbox": { "b": 287.2391357421875, "coord_origin": "BOTTOMLEFT", "l": 354.4414367675781, "r": 399.9187927246094, "t": 295.7132873535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 35, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 35, "text": "" }, { "bbox": { "b": 287.7720031738281, "coord_origin": "BOTTOMLEFT", "l": 448.2947692871094, "r": 500.7635192871094, "t": 296.0963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 36, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 35, "text": "175,098,365,045" }, { "bbox": { "b": 277.56463623046875, "coord_origin": "BOTTOMLEFT", "l": 108.71043395996094, "r": 132.465576171875, "t": 286.0724792480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 36, "text": "Swahili" }, { "bbox": { "b": 277.816162109375, "coord_origin": "BOTTOMLEFT", "l": 204.20738220214844, "r": 216.68763732910156, "t": 286.2982177734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 36, "text": "swh" }, { "bbox": { "b": 278.0283203125, "coord_origin": "BOTTOMLEFT", "l": 245.92393493652344, "r": 253.94558715820312, "t": 286.57135009765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 36, "text": "sw" }, { "bbox": { "b": 278.00048828125, "coord_origin": "BOTTOMLEFT", "l": 271.4209899902344, "r": 300.7360534667969, "t": 286.3871154785156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 36, "text": "Bantoid" }, { "bbox": { "b": 277.2432861328125, "coord_origin": "BOTTOMLEFT", "l": 354.04034423828125, "r": 393.48223876953125, "t": 285.7327880859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 36, "text": "Niger-Congo" }, { "bbox": { "b": 277.5732727050781, "coord_origin": "BOTTOMLEFT", "l": 406.56268310546875, "r": 429.0010986328125, "t": 286.0862731933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 36, "text": "Africa" }, { "bbox": { "b": 277.6167907714844, "coord_origin": "BOTTOMLEFT", "l": 456.8694763183594, "r": 501.2992248535156, "t": 285.93365478515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 37, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 36, "text": "236,482,543" }, { "bbox": { "b": 266.7707214355469, "coord_origin": "BOTTOMLEFT", "l": 108.87135314941406, "r": 130.01206970214844, "t": 275.24664306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 37, "text": "Tamil" }, { "bbox": { "b": 267.4174499511719, "coord_origin": "BOTTOMLEFT", "l": 203.4427032470703, "r": 217.20338439941406, "t": 275.8340148925781 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 37, "text": "tam" }, { "bbox": { "b": 267.2631530761719, "coord_origin": "BOTTOMLEFT", "l": 246.16098022460938, "r": 253.42066955566406, "t": 275.7876281738281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 37, "text": "ta" }, { "bbox": { "b": 266.78253173828125, "coord_origin": "BOTTOMLEFT", "l": 273.2210693359375, "r": 339.6412353515625, "t": 275.2291259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 37, "text": "Southern Dravidian" }, { "bbox": { "b": 266.7969665527344, "coord_origin": "BOTTOMLEFT", "l": 354.81964111328125, "r": 383.95281982421875, "t": 275.200439453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 37, "text": "Dravidian" }, { "bbox": { "b": 267.22747802734375, "coord_origin": "BOTTOMLEFT", "l": 406.36419677734375, "r": 429.8748474121094, "t": 275.7294616699219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 37, "text": "Eurasia" }, { "bbox": { "b": 267.181640625, "coord_origin": "BOTTOMLEFT", "l": 456.1546936035156, "r": 501.1959533691406, "t": 275.45147705078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 38, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 37, "text": "7,989,206,220" }, { "bbox": { "b": 257.2725524902344, "coord_origin": "BOTTOMLEFT", "l": 108.70472717285156, "r": 128.7847442626953, "t": 265.7151794433594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 38, "text": "Telugu" }, { "bbox": { "b": 256.9312438964844, "coord_origin": "BOTTOMLEFT", "l": 204.9400634765625, "r": 215.74571228027344, "t": 265.36444091796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 38, "text": "tel" }, { "bbox": { "b": 257.34429931640625, "coord_origin": "BOTTOMLEFT", "l": 246.07130432128906, "r": 253.7733154296875, "t": 265.8937683105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 38, "text": "te" }, { "bbox": { "b": 257.2600402832031, "coord_origin": "BOTTOMLEFT", "l": 273.256103515625, "r": 346.8114929199219, "t": 265.7115478515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 38, "text": "South-Central Dravidian Dravidian" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 38, "text": "" }, { "bbox": { "b": 257.0427551269531, "coord_origin": "BOTTOMLEFT", "l": 406.3435974121094, "r": 429.6686096191406, "t": 265.54815673828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 38, "text": "Eurasia" }, { "bbox": { "b": 257.5414733886719, "coord_origin": "BOTTOMLEFT", "l": 460.3102111816406, "r": 501.2451171875, "t": 265.8232421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 39, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 38, "text": "2993407,159" }, { "bbox": { "b": 246.57666015625, "coord_origin": "BOTTOMLEFT", "l": 108.58483123779297, "r": 171.8043975830078, "t": 254.88795471191406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 39, "text": "Traditional Chinese" }, { "bbox": { "b": 246.85757446289062, "coord_origin": "BOTTOMLEFT", "l": 205.56141662597656, "r": 214.78268432617188, "t": 255.35275268554688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 39, "text": "-" }, { "bbox": { "b": 246.96266174316406, "coord_origin": "BOTTOMLEFT", "l": 245.1876983642578, "r": 254.34671020507812, "t": 255.5424041748047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 39, "text": "zht" }, { "bbox": { "b": 247.37074279785156, "coord_origin": "BOTTOMLEFT", "l": 272.2436828613281, "r": 303.81634521484375, "t": 255.70603942871094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 39, "text": "Chinese" }, { "bbox": { "b": 247.2266082763672, "coord_origin": "BOTTOMLEFT", "l": 355.0390625, "r": 393.97882080078125, "t": 255.6996307373047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 39, "text": "Sino-Tibetan" }, { "bbox": { "b": 246.9362335205078, "coord_origin": "BOTTOMLEFT", "l": 406.3245849609375, "r": 429.05096435546875, "t": 255.42367553710938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 39, "text": "Eurasia" }, { "bbox": { "b": 246.86868286132812, "coord_origin": "BOTTOMLEFT", "l": 463.15142822265625, "r": 501.22027587890625, "t": 255.15745544433594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 40, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 39, "text": "762,489,150" }, { "bbox": { "b": 236.9476318359375, "coord_origin": "BOTTOMLEFT", "l": 108.84326934814453, "r": 123.63748168945312, "t": 245.3838653564453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 40, "text": "Twi" }, { "bbox": { "b": 236.7466583251953, "coord_origin": "BOTTOMLEFT", "l": 205.5863037109375, "r": 215.42103576660156, "t": 245.21051025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 40, "text": "twi" }, { "bbox": { "b": 237.1617889404297, "coord_origin": "BOTTOMLEFT", "l": 246.14259338378906, "r": 253.93890380859375, "t": 245.7530059814453 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 40, "text": "tw" }, { "bbox": { "b": 237.04428100585938, "coord_origin": "BOTTOMLEFT", "l": 271.3865966796875, "r": 286.4140319824219, "t": 245.48175048828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 40, "text": "Kwa" }, { "bbox": { "b": 236.35682678222656, "coord_origin": "BOTTOMLEFT", "l": 354.77606201171875, "r": 398.3659973144531, "t": 244.86277770996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 40, "text": "Niger-Congo" }, { "bbox": { "b": 236.8165740966797, "coord_origin": "BOTTOMLEFT", "l": 406.5733947753906, "r": 428.7755432128906, "t": 245.31649780273438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 40, "text": "Africa" }, { "bbox": { "b": 236.84291076660156, "coord_origin": "BOTTOMLEFT", "l": 467.62652587890625, "r": 500.6257019042969, "t": 245.17544555664062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 41, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 40, "text": "1,265,041" }, { "bbox": { "b": 227.35032653808594, "coord_origin": "BOTTOMLEFT", "l": 108.52404022216797, "r": 124.53902435302734, "t": 235.76144409179688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 41, "text": "Urdu" }, { "bbox": { "b": 227.0521240234375, "coord_origin": "BOTTOMLEFT", "l": 205.2458953857422, "r": 216.35084533691406, "t": 235.49473571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 41, "text": "urd" }, { "bbox": { "b": 227.05445861816406, "coord_origin": "BOTTOMLEFT", "l": 246.3316192626953, "r": 253.7119903564453, "t": 235.62001037597656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 41, "text": "ur" }, { "bbox": { "b": 227.7138214111328, "coord_origin": "BOTTOMLEFT", "l": 271.7252197265625, "r": 284.8417053222656, "t": 236.22555541992188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 41, "text": "Indic" }, { "bbox": { "b": 226.95724487304688, "coord_origin": "BOTTOMLEFT", "l": 354.75848388671875, "r": 402.03424072265625, "t": 235.4998016357422 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 41, "text": "Indo-European Eurasia" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 41, "text": "" }, { "bbox": { "b": 227.02642822265625, "coord_origin": "BOTTOMLEFT", "l": 454.41302490234375, "r": 500.4913635253906, "t": 235.41854858398438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 42, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 41, "text": "2,781,329,959" }, { "bbox": { "b": 216.3731689453125, "coord_origin": "BOTTOMLEFT", "l": 108.97728729248047, "r": 144.2902069091797, "t": 224.86988830566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 42, "text": "Vietnamese" }, { "bbox": { "b": 216.7290802001953, "coord_origin": "BOTTOMLEFT", "l": 205.62844848632812, "r": 216.28665161132812, "t": 225.19895935058594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 42, "text": "vie" }, { "bbox": { "b": 216.7925567626953, "coord_origin": "BOTTOMLEFT", "l": 246.4274139404297, "r": 253.82920837402344, "t": 225.45684814453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 42, "text": "vi" }, { "bbox": { "b": 217.48045349121094, "coord_origin": "BOTTOMLEFT", "l": 271.5227966308594, "r": 308.4107971191406, "t": 225.9589080810547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 42, "text": "Viet-Muong" }, { "bbox": { "b": 216.5958709716797, "coord_origin": "BOTTOMLEFT", "l": 354.7583923339844, "r": 401.1172180175781, "t": 225.18240356445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 42, "text": "Austro-Asiatic" }, { "bbox": { "b": 216.71151733398438, "coord_origin": "BOTTOMLEFT", "l": 406.1020812988281, "r": 429.02044677734375, "t": 225.3156280517578 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 42, "text": "Eurasia" }, { "bbox": { "b": 217.2748565673828, "coord_origin": "BOTTOMLEFT", "l": 456.14923095703125, "r": 500.35308837890625, "t": 225.66273498535156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 43, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 42, "text": "43,709,279,959" }, { "bbox": { "b": 206.9718475341797, "coord_origin": "BOTTOMLEFT", "l": 108.85558319091797, "r": 129.42886352539062, "t": 215.3903350830078 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 43, "text": "Wolof" }, { "bbox": { "b": 206.7169647216797, "coord_origin": "BOTTOMLEFT", "l": 205.2184295654297, "r": 216.8064422607422, "t": 215.29913330078125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 43, "text": "wol" }, { "bbox": { "b": 206.982421875, "coord_origin": "BOTTOMLEFT", "l": 245.613037109375, "r": 254.1932830810547, "t": 215.71926879882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 43, "text": "wo" }, { "bbox": { "b": 207.22157287597656, "coord_origin": "BOTTOMLEFT", "l": 270.56488037109375, "r": 290.053466796875, "t": 215.87628173828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 43, "text": "Wolof" }, { "bbox": { "b": 206.91802978515625, "coord_origin": "BOTTOMLEFT", "l": 354.80926513671875, "r": 398.0904235839844, "t": 215.5409393310547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 43, "text": "Niger-Congo" }, { "bbox": { "b": 206.661865234375, "coord_origin": "BOTTOMLEFT", "l": 406.0134582519531, "r": 428.2970886230469, "t": 215.3527069091797 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 43, "text": "Africa" }, { "bbox": { "b": 206.67898559570312, "coord_origin": "BOTTOMLEFT", "l": 466.620849609375, "r": 500.9853820800781, "t": 215.0123291015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 44, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 43, "text": "3,606,973" }, { "bbox": { "b": 196.03115844726562, "coord_origin": "BOTTOMLEFT", "l": 108.87499237060547, "r": 130.56085205078125, "t": 204.55064392089844 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 44, "text": "Xhosa" }, { "bbox": { "b": 196.32159423828125, "coord_origin": "BOTTOMLEFT", "l": 204.9607696533203, "r": 216.56484985351562, "t": 204.9009246826172 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 44, "text": "xho" }, { "bbox": { "b": 196.44464111328125, "coord_origin": "BOTTOMLEFT", "l": 246.29147338867188, "r": 253.8073272705078, "t": 205.17361450195312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 44, "text": "xh" }, { "bbox": { "b": 195.97079467773438, "coord_origin": "BOTTOMLEFT", "l": 271.2140197753906, "r": 291.8352355957031, "t": 204.65423583984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 44, "text": "Bantoid" }, { "bbox": { "b": 195.67611694335938, "coord_origin": "BOTTOMLEFT", "l": 355.1220703125, "r": 396.5032043457031, "t": 204.26348876953125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 44, "text": "Niger-Congo" }, { "bbox": { "b": 195.8412628173828, "coord_origin": "BOTTOMLEFT", "l": 405.8934020996094, "r": 427.184814453125, "t": 204.533935546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 44, "text": "Africa" }, { "bbox": { "b": 195.99139404296875, "coord_origin": "BOTTOMLEFT", "l": 466.9107971191406, "r": 500.642822265625, "t": 204.3134002685547 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 45, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 44, "text": "14,304,074" }, { "bbox": { "b": 185.78851318359375, "coord_origin": "BOTTOMLEFT", "l": 108.67925262451172, "r": 134.6645050048828, "t": 194.46615600585938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 45, "text": "Xitsonga" }, { "bbox": { "b": 186.1658935546875, "coord_origin": "BOTTOMLEFT", "l": 205.25071716308594, "r": 215.73153686523438, "t": 194.8161163330078 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 45, "text": "tso" }, { "bbox": { "b": 186.6497039794922, "coord_origin": "BOTTOMLEFT", "l": 246.6892852783203, "r": 253.50070190429688, "t": 195.4151611328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 45, "text": "ts" }, { "bbox": { "b": 186.28427124023438, "coord_origin": "BOTTOMLEFT", "l": 272.11260986328125, "r": 294.1134338378906, "t": 194.98060607910156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 45, "text": "Bantoid" }, { "bbox": { "b": 186.032470703125, "coord_origin": "BOTTOMLEFT", "l": 355.0280456542969, "r": 396.8128967285156, "t": 194.68316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 45, "text": "Niger-Congo" }, { "bbox": { "b": 186.06903076171875, "coord_origin": "BOTTOMLEFT", "l": 406.20843505859375, "r": 426.8867492675781, "t": 194.7771453857422 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 45, "text": "Africa" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 46, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 45, "text": "" }, { "bbox": { "b": 175.14283752441406, "coord_origin": "BOTTOMLEFT", "l": 108.71174621582031, "r": 139.44134521484375, "t": 183.88453674316406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 46, "text": "Yoruba" }, { "bbox": { "b": 174.76832580566406, "coord_origin": "BOTTOMLEFT", "l": 205.20335388183594, "r": 214.9654998779297, "t": 183.44154357910156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 46, "text": "yor" }, { "bbox": { "b": 175.76734924316406, "coord_origin": "BOTTOMLEFT", "l": 246.59974670410156, "r": 253.18612670898438, "t": 184.48805236816406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 46, "text": "yo" }, { "bbox": { "b": 175.52835083007812, "coord_origin": "BOTTOMLEFT", "l": 272.08477783203125, "r": 293.53045654296875, "t": 184.17129516601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 46, "text": "Defoid" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 46, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 46, "text": "" }, { "bbox": { "b": 183.8022003173828, "coord_origin": "BOTTOMLEFT", "l": 469.1238708496094, "r": 500.5533142089844, "t": 192.45419311523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 47, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 46, "text": "707,634" }, { "bbox": { "b": 163.75592041015625, "coord_origin": "BOTTOMLEFT", "l": 108.24192810058594, "r": 187.58399963378906, "t": 172.73565673828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 47, "text": "Programming Languages" }, { "bbox": { "b": 164.87973022460938, "coord_origin": "BOTTOMLEFT", "l": 204.60964965820312, "r": 214.92276000976562, "t": 173.8241729736328 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 47, "text": "-" }, { "bbox": { "b": 165.4375457763672, "coord_origin": "BOTTOMLEFT", "l": 246.6812286376953, "r": 253.26661682128906, "t": 174.3863525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 47, "text": "-" }, { "bbox": { "b": 167.28050231933594, "coord_origin": "BOTTOMLEFT", "l": 271.4575500488281, "r": 291.3991394042969, "t": 176.1390838623047 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 47, "text": "-" }, { "bbox": { "b": 164.84762573242188, "coord_origin": "BOTTOMLEFT", "l": 355.00469970703125, "r": 395.3666687011719, "t": 183.07347106933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 47, "text": "Niger-Congo -" }, { "bbox": { "b": 174.52589416503906, "coord_origin": "BOTTOMLEFT", "l": 406.2784423828125, "r": 426.1576843261719, "t": 183.22804260253906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 47, "text": "Africa" }, { "bbox": { "b": 164.84762573242188, "coord_origin": "BOTTOMLEFT", "l": 448.7660217285156, "r": 501.6629638671875, "t": 183.8473358154297 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 48, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 47, "text": "89,695,835 174,700,245,772" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 161.95001220703125, "coord_origin": "BOTTOMLEFT", "l": 108.18231201171875, "r": 503.786376953125, "t": 658.2073364257812 }, "charspan": [ 0, 0 ], "page_no": 32 } ], "references": [], "self_ref": "#/tables/9" }, { "captions": [ { "$ref": "#/texts/341" } ], "children": [], "data": { "grid": [ [ { "bbox": { "b": 492.0348205566406, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 196.39205932617188, "t": 500.9912109375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 492.0348205566406, "coord_origin": "BOTTOMLEFT", "l": 236.4118194580078, "r": 389.4374084472656, "t": 500.9912109375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Number of Pseudocrawled Domains" }, { "bbox": { "b": 492.0348205566406, "coord_origin": "BOTTOMLEFT", "l": 403.9927673339844, "r": 457.6812438964844, "t": 500.9912109375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Size in Bytes" } ], [ { "bbox": { "b": 473.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 185.87156677246094, "t": 482.7646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Spanish" }, { "bbox": { "b": 473.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 305.45263671875, "r": 320.39654541015625, "t": 482.7646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "108" }, { "bbox": { "b": 473.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 395.4149169921875, "r": 457.68115234375, "t": 482.7646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "29,440,210,712" } ], [ { "bbox": { "b": 460.9560852050781, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 184.76571655273438, "t": 469.8626403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "English" }, { "bbox": { "b": 460.9560852050781, "coord_origin": "BOTTOMLEFT", "l": 307.94329833984375, "r": 317.9058837890625, "t": 469.8626403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "22" }, { "bbox": { "b": 460.9560852050781, "coord_origin": "BOTTOMLEFT", "l": 400.3962097167969, "r": 457.68115234375, "t": 469.8626403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "4,537,031,408" } ], [ { "bbox": { "b": 448.0550842285156, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 184.6660919189453, "t": 456.9616394042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Swahili" }, { "bbox": { "b": 448.0550842285156, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 456.9616394042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "5" }, { "bbox": { "b": 448.0550842285156, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 456.9616394042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "109,110,002" } ], [ { "bbox": { "b": 435.153076171875, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 198.035888671875, "t": 444.05963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Indonesian" }, { "bbox": { "b": 435.153076171875, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 444.05963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "4" }, { "bbox": { "b": 435.153076171875, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 444.05963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "770,023,233" } ], [ { "bbox": { "b": 422.2520751953125, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 183.64990234375, "t": 431.15863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 422.2520751953125, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 431.15863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "4" }, { "bbox": { "b": 422.2520751953125, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 431.15863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "281,610,312" } ], [ { "bbox": { "b": 409.3500671386719, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 181.98614501953125, "t": 418.2566223144531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "French" }, { "bbox": { "b": 409.3500671386719, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 418.2566223144531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "3" }, { "bbox": { "b": 409.3500671386719, "coord_origin": "BOTTOMLEFT", "l": 400.396240234375, "r": 457.6811828613281, "t": 418.2566223144531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "1,416,682,404" } ], [ { "bbox": { "b": 396.44805908203125, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 177.01480102539062, "t": 405.3546142578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Hindi" }, { "bbox": { "b": 396.44805908203125, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 405.3546142578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "2" }, { "bbox": { "b": 396.44805908203125, "coord_origin": "BOTTOMLEFT", "l": 400.39617919921875, "r": 457.6811218261719, "t": 405.3546142578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "1,536,649,276" } ], [ { "bbox": { "b": 383.54705810546875, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 230.43426513671875, "t": 392.45361328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Simplified Chinese" }, { "bbox": { "b": 383.54705810546875, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 392.45361328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "2" }, { "bbox": { "b": 383.54705810546875, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 392.45361328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "173,884,238" } ], [ { "bbox": { "b": 370.6450500488281, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 183.10195922851562, "t": 379.5516052246094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Yoruba" }, { "bbox": { "b": 370.6450500488281, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 379.5516052246094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "2" }, { "bbox": { "b": 370.6450500488281, "coord_origin": "BOTTOMLEFT", "l": 417.8307800292969, "r": 457.6811828613281, "t": 379.5516052246094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "6,198,347" } ], [ { "bbox": { "b": 357.7440490722656, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 172.58145141601562, "t": 366.6506042480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Igbo" }, { "bbox": { "b": 357.7440490722656, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 366.6506042480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "2" }, { "bbox": { "b": 357.7440490722656, "coord_origin": "BOTTOMLEFT", "l": 417.8307800292969, "r": 457.6811828613281, "t": 366.6506042480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "2,650,116" } ], [ { "bbox": { "b": 344.842041015625, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 181.42823791503906, "t": 353.74859619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Arabic" }, { "bbox": { "b": 344.842041015625, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 353.74859619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "1" }, { "bbox": { "b": 344.842041015625, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 353.74859619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "694,455,304" } ], [ { "bbox": { "b": 331.9400329589844, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 198.5937957763672, "t": 340.8465881347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Portuguese" }, { "bbox": { "b": 331.9400329589844, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 340.8465881347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "1" }, { "bbox": { "b": 331.9400329589844, "coord_origin": "BOTTOMLEFT", "l": 412.8494567871094, "r": 457.68115234375, "t": 340.8465881347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "30,615,557" } ], [ { "bbox": { "b": 319.0390319824219, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 207.7394561767578, "t": 327.9455871582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Kinyarwanda" }, { "bbox": { "b": 319.0390319824219, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 327.9455871582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "1" }, { "bbox": { "b": 319.0390319824219, "coord_origin": "BOTTOMLEFT", "l": 417.83074951171875, "r": 457.68115234375, "t": 327.9455871582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "9,301,301" } ] ], "num_cols": 3, "num_rows": 14, "table_cells": [ { "bbox": { "b": 492.0348205566406, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 196.39205932617188, "t": 500.9912109375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Language" }, { "bbox": { "b": 492.0348205566406, "coord_origin": "BOTTOMLEFT", "l": 236.4118194580078, "r": 389.4374084472656, "t": 500.9912109375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Number of Pseudocrawled Domains" }, { "bbox": { "b": 492.0348205566406, "coord_origin": "BOTTOMLEFT", "l": 403.9927673339844, "r": 457.6812438964844, "t": 500.9912109375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Size in Bytes" }, { "bbox": { "b": 473.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 185.87156677246094, "t": 482.7646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Spanish" }, { "bbox": { "b": 473.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 305.45263671875, "r": 320.39654541015625, "t": 482.7646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "108" }, { "bbox": { "b": 473.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 395.4149169921875, "r": 457.68115234375, "t": 482.7646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "29,440,210,712" }, { "bbox": { "b": 460.9560852050781, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 184.76571655273438, "t": 469.8626403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "English" }, { "bbox": { "b": 460.9560852050781, "coord_origin": "BOTTOMLEFT", "l": 307.94329833984375, "r": 317.9058837890625, "t": 469.8626403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "22" }, { "bbox": { "b": 460.9560852050781, "coord_origin": "BOTTOMLEFT", "l": 400.3962097167969, "r": 457.68115234375, "t": 469.8626403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "4,537,031,408" }, { "bbox": { "b": 448.0550842285156, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 184.6660919189453, "t": 456.9616394042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Swahili" }, { "bbox": { "b": 448.0550842285156, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 456.9616394042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "5" }, { "bbox": { "b": 448.0550842285156, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 456.9616394042969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "109,110,002" }, { "bbox": { "b": 435.153076171875, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 198.035888671875, "t": 444.05963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Indonesian" }, { "bbox": { "b": 435.153076171875, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 444.05963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "4" }, { "bbox": { "b": 435.153076171875, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 444.05963134765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "770,023,233" }, { "bbox": { "b": 422.2520751953125, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 183.64990234375, "t": 431.15863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Basque" }, { "bbox": { "b": 422.2520751953125, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 431.15863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "4" }, { "bbox": { "b": 422.2520751953125, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 431.15863037109375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "281,610,312" }, { "bbox": { "b": 409.3500671386719, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 181.98614501953125, "t": 418.2566223144531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "French" }, { "bbox": { "b": 409.3500671386719, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 418.2566223144531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "3" }, { "bbox": { "b": 409.3500671386719, "coord_origin": "BOTTOMLEFT", "l": 400.396240234375, "r": 457.6811828613281, "t": 418.2566223144531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "1,416,682,404" }, { "bbox": { "b": 396.44805908203125, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 177.01480102539062, "t": 405.3546142578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Hindi" }, { "bbox": { "b": 396.44805908203125, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 405.3546142578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "2" }, { "bbox": { "b": 396.44805908203125, "coord_origin": "BOTTOMLEFT", "l": 400.39617919921875, "r": 457.6811218261719, "t": 405.3546142578125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "1,536,649,276" }, { "bbox": { "b": 383.54705810546875, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 230.43426513671875, "t": 392.45361328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Simplified Chinese" }, { "bbox": { "b": 383.54705810546875, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 392.45361328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "2" }, { "bbox": { "b": 383.54705810546875, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 392.45361328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "173,884,238" }, { "bbox": { "b": 370.6450500488281, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 183.10195922851562, "t": 379.5516052246094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Yoruba" }, { "bbox": { "b": 370.6450500488281, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 379.5516052246094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "2" }, { "bbox": { "b": 370.6450500488281, "coord_origin": "BOTTOMLEFT", "l": 417.8307800292969, "r": 457.6811828613281, "t": 379.5516052246094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "6,198,347" }, { "bbox": { "b": 357.7440490722656, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 172.58145141601562, "t": 366.6506042480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Igbo" }, { "bbox": { "b": 357.7440490722656, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 366.6506042480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "2" }, { "bbox": { "b": 357.7440490722656, "coord_origin": "BOTTOMLEFT", "l": 417.8307800292969, "r": 457.6811828613281, "t": 366.6506042480469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "2,650,116" }, { "bbox": { "b": 344.842041015625, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 181.42823791503906, "t": 353.74859619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Arabic" }, { "bbox": { "b": 344.842041015625, "coord_origin": "BOTTOMLEFT", "l": 310.4339599609375, "r": 315.4152526855469, "t": 353.74859619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "1" }, { "bbox": { "b": 344.842041015625, "coord_origin": "BOTTOMLEFT", "l": 407.8681640625, "r": 457.68115234375, "t": 353.74859619140625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "694,455,304" }, { "bbox": { "b": 331.9400329589844, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 198.5937957763672, "t": 340.8465881347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Portuguese" }, { "bbox": { "b": 331.9400329589844, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 340.8465881347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "1" }, { "bbox": { "b": 331.9400329589844, "coord_origin": "BOTTOMLEFT", "l": 412.8494567871094, "r": 457.68115234375, "t": 340.8465881347656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "30,615,557" }, { "bbox": { "b": 319.0390319824219, "coord_origin": "BOTTOMLEFT", "l": 154.32000732421875, "r": 207.7394561767578, "t": 327.9455871582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "Kinyarwanda" }, { "bbox": { "b": 319.0390319824219, "coord_origin": "BOTTOMLEFT", "l": 310.4339294433594, "r": 315.41522216796875, "t": 327.9455871582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "1" }, { "bbox": { "b": 319.0390319824219, "coord_origin": "BOTTOMLEFT", "l": 417.83074951171875, "r": 457.68115234375, "t": 327.9455871582031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "9,301,301" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 315.00311279296875, "coord_origin": "BOTTOMLEFT", "l": 151.10562133789062, "r": 460.97222900390625, "t": 508.32012939453125 }, "charspan": [ 0, 0 ], "page_no": 33 } ], "references": [], "self_ref": "#/tables/10" } ]

[]

{ "1": { "page_no": 1, "size": { "height": 792, "width": 612 } }, "2": { "page_no": 2, "size": { "height": 792, "width": 612 } }, "3": { "page_no": 3, "size": { "height": 792, "width": 612 } }, "4": { "page_no": 4, "size": { "height": 792, "width": 612 } }, "5": { "page_no": 5, "size": { "height": 792, "width": 612 } }, "6": { "page_no": 6, "size": { "height": 792, "width": 612 } }, "7": { "page_no": 7, "size": { "height": 792, "width": 612 } }, "8": { "page_no": 8, "size": { "height": 792, "width": 612 } }, "9": { "page_no": 9, "size": { "height": 792, "width": 612 } }, "10": { "page_no": 10, "size": { "height": 792, "width": 612 } }, "11": { "page_no": 11, "size": { "height": 792, "width": 612 } }, "12": { "page_no": 12, "size": { "height": 792, "width": 612 } }, "13": { "page_no": 13, "size": { "height": 792, "width": 612 } }, "14": { "page_no": 14, "size": { "height": 792, "width": 612 } }, "15": { "page_no": 15, "size": { "height": 792, "width": 612 } }, "16": { "page_no": 16, "size": { "height": 792, "width": 612 } }, "17": { "page_no": 17, "size": { "height": 792, "width": 612 } }, "18": { "page_no": 18, "size": { "height": 792, "width": 612 } }, "19": { "page_no": 19, "size": { "height": 792, "width": 612 } }, "20": { "page_no": 20, "size": { "height": 792, "width": 612 } }, "21": { "page_no": 21, "size": { "height": 792, "width": 612 } }, "22": { "page_no": 22, "size": { "height": 792, "width": 612 } }, "23": { "page_no": 23, "size": { "height": 792, "width": 612 } }, "24": { "page_no": 24, "size": { "height": 792, "width": 612 } }, "25": { "page_no": 25, "size": { "height": 792, "width": 612 } }, "26": { "page_no": 26, "size": { "height": 792, "width": 612 } }, "27": { "page_no": 27, "size": { "height": 792, "width": 612 } }, "28": { "page_no": 28, "size": { "height": 792, "width": 612 } }, "29": { "page_no": 29, "size": { "height": 792, "width": 612 } }, "30": { "page_no": 30, "size": { "height": 792, "width": 612 } }, "31": { "page_no": 31, "size": { "height": 792, "width": 612 } }, "32": { "page_no": 32, "size": { "height": 792, "width": 612 } }, "33": { "page_no": 33, "size": { "height": 792, "width": 612 } }, "34": { "page_no": 34, "size": { "height": 792, "width": 612 } }, "35": { "page_no": 35, "size": { "height": 792, "width": 612 } }, "36": { "page_no": 36, "size": { "height": 792, "width": 612 } } }

{ "children": [], "label": "unspecified", "name": "_root_", "self_ref": "#/furniture" }

{ "children": [ { "$ref": "#/texts/0" }, { "$ref": "#/pictures/0" }, { "$ref": "#/texts/1" }, { "$ref": "#/texts/2" }, { "$ref": "#/texts/3" }, { "$ref": "#/texts/4" }, { "$ref": "#/texts/5" }, { "$ref": "#/texts/6" }, { "$ref": "#/texts/7" }, { "$ref": "#/texts/8" }, { "$ref": "#/texts/9" }, { "$ref": "#/texts/10" }, { "$ref": "#/texts/11" }, { "$ref": "#/groups/0" }, { "$ref": "#/texts/13" }, { "$ref": "#/groups/1" }, { "$ref": "#/texts/16" }, { "$ref": "#/texts/17" }, { "$ref": "#/texts/18" }, { "$ref": "#/texts/19" }, { "$ref": "#/texts/20" }, { "$ref": "#/texts/21" }, { "$ref": "#/texts/22" }, { "$ref": "#/texts/23" }, { "$ref": "#/texts/24" }, { "$ref": "#/texts/25" }, { "$ref": "#/texts/26" }, { "$ref": "#/texts/27" }, { "$ref": "#/texts/28" }, { "$ref": "#/texts/29" }, { "$ref": "#/texts/30" }, { "$ref": "#/texts/31" }, { "$ref": "#/texts/32" }, { "$ref": "#/texts/33" }, { "$ref": "#/groups/2" }, { "$ref": "#/texts/36" }, { "$ref": "#/texts/37" }, { "$ref": "#/texts/38" }, { "$ref": "#/texts/39" }, { "$ref": "#/texts/40" }, { "$ref": "#/texts/41" }, { "$ref": "#/texts/42" }, { "$ref": "#/texts/43" }, { "$ref": "#/texts/44" }, { "$ref": "#/texts/45" }, { "$ref": "#/pictures/1" }, { "$ref": "#/texts/46" }, { "$ref": "#/texts/47" }, { "$ref": "#/texts/48" }, { "$ref": "#/texts/49" }, { "$ref": "#/texts/50" }, { "$ref": "#/texts/51" }, { "$ref": "#/texts/52" }, { "$ref": "#/texts/53" }, { "$ref": "#/texts/54" }, { "$ref": "#/texts/55" }, { "$ref": "#/texts/56" }, { "$ref": "#/texts/57" }, { "$ref": "#/texts/58" }, { "$ref": "#/texts/59" }, { "$ref": "#/texts/60" }, { "$ref": "#/texts/61" }, { "$ref": "#/texts/62" }, { "$ref": "#/texts/63" }, { "$ref": "#/texts/64" }, { "$ref": "#/texts/65" }, { "$ref": "#/pictures/2" }, { "$ref": "#/texts/66" }, { "$ref": "#/texts/67" }, { "$ref": "#/texts/68" }, { "$ref": "#/texts/69" }, { "$ref": "#/texts/70" }, { "$ref": "#/texts/71" }, { "$ref": "#/texts/72" }, { "$ref": "#/texts/73" }, { "$ref": "#/texts/74" }, { "$ref": "#/pictures/3" }, { "$ref": "#/texts/75" }, { "$ref": "#/texts/76" }, { "$ref": "#/texts/77" }, { "$ref": "#/texts/78" }, { "$ref": "#/texts/79" }, { "$ref": "#/texts/80" }, { "$ref": "#/texts/81" }, { "$ref": "#/texts/82" }, { "$ref": "#/texts/83" }, { "$ref": "#/texts/84" }, { "$ref": "#/texts/85" }, { "$ref": "#/texts/86" }, { "$ref": "#/texts/87" }, { "$ref": "#/groups/3" }, { "$ref": "#/texts/89" }, { "$ref": "#/texts/90" }, { "$ref": "#/texts/91" }, { "$ref": "#/texts/92" }, { "$ref": "#/texts/93" }, { "$ref": "#/texts/94" }, { "$ref": "#/texts/95" }, { "$ref": "#/texts/96" }, { "$ref": "#/texts/97" }, { "$ref": "#/texts/98" }, { "$ref": "#/texts/99" }, { "$ref": "#/texts/100" }, { "$ref": "#/texts/101" }, { "$ref": "#/texts/102" }, { "$ref": "#/texts/103" }, { "$ref": "#/texts/104" }, { "$ref": "#/texts/105" }, { "$ref": "#/texts/106" }, { "$ref": "#/texts/107" }, { "$ref": "#/texts/108" }, { "$ref": "#/texts/109" }, { "$ref": "#/texts/110" }, { "$ref": "#/texts/111" }, { "$ref": "#/texts/112" }, { "$ref": "#/texts/113" }, { "$ref": "#/texts/114" }, { "$ref": "#/texts/115" }, { "$ref": "#/texts/116" }, { "$ref": "#/texts/117" }, { "$ref": "#/texts/118" }, { "$ref": "#/texts/119" }, { "$ref": "#/texts/120" }, { "$ref": "#/texts/121" }, { "$ref": "#/texts/122" } ], "label": "unspecified", "name": "_root_", "self_ref": "#/body" }

[ { "children": [ { "$ref": "#/texts/12" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/0" }, { "children": [ { "$ref": "#/texts/14" }, { "$ref": "#/texts/15" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/1" }, { "children": [ { "$ref": "#/texts/34" }, { "$ref": "#/texts/35" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/2" }, { "children": [ { "$ref": "#/texts/88" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/3" } ]

[ { "children": [], "enumerated": null, "label": "page_header", "level": null, "marker": null, "orig": "arXiv:2109.02846v1 [cs.CL] 7 Sep 2021", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 236.99996948242188, "coord_origin": "BOTTOMLEFT", "l": 17.535791397094727, "r": 36.33979415893555, "t": 573.6400146484375 }, "charspan": [ 0, 37 ], "page_no": 1 } ], "self_ref": "#/texts/0", "text": "arXiv:2109.02846v1 [cs.CL] 7 Sep 2021" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Datasets: A Community Library for Natural Language Processing", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 754.0599975585938, "coord_origin": "BOTTOMLEFT", "l": 108.30389404296875, "r": 515.3463745117188, "t": 768.6941528320312 }, "charspan": [ 0, 61 ], "page_no": 1 } ], "self_ref": "#/texts/1", "text": "Datasets: A Community Library for Natural Language Processing" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Quentin Lhoest $^{∗}$, Albert Villanova del Moral$^{∗}$, Yacine Jernite, Abhishek Thakur, Patrick von Platen, Suraj Patil, Julien Chaumond, Mariama Drame, Julien Plu, Lewis Tunstall, Joe Davison, Mario Šaško $^{↑}$, Gunjan Chhablani$^{↑}$, Bhavitvya Malik$^{↑}$, Simon Brandeis, Teven Le Scao, Victor Sanh, Canwen Xu, Nicolas Patry, Angelina McMillan-Major, Philipp Schmid, Sylvain Gugger, Clément Delangue, Théo Matussière, Lysandre Debut, Stas Bekman, Pierric Cistac, Thibault Goehringer, Victor Mustar, François Lagunas, Alexander M. Rush, and Thomas Wolf Ω", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 636.0336303710938, "coord_origin": "BOTTOMLEFT", "l": 72.90849304199219, "r": 524.9769897460938, "t": 732.2125244140625 }, "charspan": [ 0, 560 ], "page_no": 1 } ], "self_ref": "#/texts/2", "text": "Quentin Lhoest $^{∗}$, Albert Villanova del Moral$^{∗}$, Yacine Jernite, Abhishek Thakur, Patrick von Platen, Suraj Patil, Julien Chaumond, Mariama Drame, Julien Plu, Lewis Tunstall, Joe Davison, Mario Šaško $^{↑}$, Gunjan Chhablani$^{↑}$, Bhavitvya Malik$^{↑}$, Simon Brandeis, Teven Le Scao, Victor Sanh, Canwen Xu, Nicolas Patry, Angelina McMillan-Major, Philipp Schmid, Sylvain Gugger, Clément Delangue, Théo Matussière, Lysandre Debut, Stas Bekman, Pierric Cistac, Thibault Goehringer, Victor Mustar, François Lagunas, Alexander M. Rush, and Thomas Wolf Ω" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hugging Face / {quentin,thomas}@huggingface.co", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 607.5208129882812, "coord_origin": "BOTTOMLEFT", "l": 175.85549926757812, "r": 421.96148681640625, "t": 619.448974609375 }, "charspan": [ 0, 46 ], "page_no": 1 } ], "self_ref": "#/texts/3", "text": "Hugging Face / {quentin,thomas}@huggingface.co" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Abstract", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 578.9833374023438, "coord_origin": "BOTTOMLEFT", "l": 156.71885681152344, "r": 202.5477752685547, "t": 590.0819091796875 }, "charspan": [ 0, 8 ], "page_no": 1 } ], "self_ref": "#/texts/4", "text": "Abstract" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The scale, variety, and quantity of publiclyavailable NLP datasets has grown rapidly as researchers propose new tasks, larger models, and novel benchmarks. Datasets is a community library for contemporary NLP designed to support this ecosystem. Datasets aims to standardize end-user interfaces, versioning, and documentation, while providing a lightweight front-end that behaves similarly for small datasets as for internet-scale corpora. The design of the library incorporates a distributed, community-driven approach to adding datasets and documenting usage. After a year of development, the library now includes more than 650 unique datasets, has more than 250 contributors, and has helped support a variety of novel crossdataset research projects and shared tasks. The library is available at https://github. com/huggingface/datasets .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 330.7523193359375, "coord_origin": "BOTTOMLEFT", "l": 86.6174545288086, "r": 276.30938720703125, "t": 568.3016357421875 }, "charspan": [ 0, 839 ], "page_no": 1 } ], "self_ref": "#/texts/5", "text": "The scale, variety, and quantity of publiclyavailable NLP datasets has grown rapidly as researchers propose new tasks, larger models, and novel benchmarks. Datasets is a community library for contemporary NLP designed to support this ecosystem. Datasets aims to standardize end-user interfaces, versioning, and documentation, while providing a lightweight front-end that behaves similarly for small datasets as for internet-scale corpora. The design of the library incorporates a distributed, community-driven approach to adding datasets and documenting usage. After a year of development, the library now includes more than 650 unique datasets, has more than 250 contributors, and has helped support a variety of novel crossdataset research projects and shared tasks. The library is available at https://github. com/huggingface/datasets ." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "1 Introduction", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 307.74749755859375, "coord_origin": "BOTTOMLEFT", "l": 70.6673583984375, "r": 153.67965698242188, "t": 318.918701171875 }, "charspan": [ 0, 14 ], "page_no": 1 } ], "self_ref": "#/texts/6", "text": "1 Introduction" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets are central to empirical NLP: curated datasets are used for evaluation and benchmarks; supervised datasets are used to train and fine-tune models; and large unsupervised datasets are necessary for pretraining and language modeling. Each dataset type differs in scale, granularity and structure, in addition to annotation methodology. Historically, new dataset paradigms have been crucial for driving the development of NLP, from the Hansard corpus for statistical machine translation (Brown et al., 1988) to the Penn Treebank for syntactic modeling (Marcus et al., 1993) to projects like OPUS and Universal Dependencies (Nivre et al., 2016; Tiedemann and Nygaard, 2004) which bring together cross-lingual data and annotations.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 96.7821044921875, "coord_origin": "BOTTOMLEFT", "l": 69.58604431152344, "r": 290.9492492675781, "t": 297.47662353515625 }, "charspan": [ 0, 735 ], "page_no": 1 } ], "self_ref": "#/texts/7", "text": "Datasets are central to empirical NLP: curated datasets are used for evaluation and benchmarks; supervised datasets are used to train and fine-tune models; and large unsupervised datasets are necessary for pretraining and language modeling. Each dataset type differs in scale, granularity and structure, in addition to annotation methodology. Historically, new dataset paradigms have been crucial for driving the development of NLP, from the Hansard corpus for statistical machine translation (Brown et al., 1988) to the Penn Treebank for syntactic modeling (Marcus et al., 1993) to projects like OPUS and Universal Dependencies (Nivre et al., 2016; Tiedemann and Nygaard, 2004) which bring together cross-lingual data and annotations." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{∗}$Lead Library Maintainers, Ω Library Creator, ↑ Independent Research Contributor", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.92925262451172, "coord_origin": "BOTTOMLEFT", "l": 69.9247055053711, "r": 290.62225341796875, "t": 89.12120056152344 }, "charspan": [ 0, 85 ], "page_no": 1 } ], "self_ref": "#/texts/8", "text": "$^{∗}$Lead Library Maintainers, Ω Library Creator, ↑ Independent Research Contributor" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Contemporary NLP systems are now developed with a pipeline that utilizes many different datasets at significantly varying scale and level of annotation (Peters et al., 2018). Different datasets are used for pretraining, fine-tuning, and benchmarking. As such, there has been a large increase in the number of datasets utilized in the NLP community. These include both large text collections like C4 (Raffel et al., 2020), fine-tuning datasets like SQuAD (Rajpurkar et al., 2016), and even complex zero-shot challenge tasks. Benchmark datasets like GLUE have been central to quantifying the the advances of models such as BERT (Wang et al., 2018; Devlin et al., 2019).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 402.98565673828125, "coord_origin": "BOTTOMLEFT", "l": 304.9332580566406, "r": 526.2255249023438, "t": 589.5250854492188 }, "charspan": [ 0, 667 ], "page_no": 1 } ], "self_ref": "#/texts/9", "text": "Contemporary NLP systems are now developed with a pipeline that utilizes many different datasets at significantly varying scale and level of annotation (Peters et al., 2018). Different datasets are used for pretraining, fine-tuning, and benchmarking. As such, there has been a large increase in the number of datasets utilized in the NLP community. These include both large text collections like C4 (Raffel et al., 2020), fine-tuning datasets like SQuAD (Rajpurkar et al., 2016), and even complex zero-shot challenge tasks. Benchmark datasets like GLUE have been central to quantifying the the advances of models such as BERT (Wang et al., 2018; Devlin et al., 2019)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The growth in datasets also brings significant challenges, including interface standardization, versioning, and documentation. A practitioner should be able to utilize N different datasets without requiring N different interfaces. In addition, N practitioners using the same dataset should know they have exactly the same version. Datasets have also grown larger, and ideally interfaces should not have to change due to this scale, whether one is using small-scale datasets like Climate Fever ( ∼ 1k data points), medium-scale Yahoo Answers ( ∼ 1M), or even all of PubMed ( ∼ 79B). Finally, datasets are being created with a variety of different procedures, from crowd-sourcing to scraping to synthetic generation, which need to be taken into account when evaluating which is most appropriate for a given purpose and ought to be immediately apparent to prospective users (Gebru et al., 2018).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 158.41314697265625, "coord_origin": "BOTTOMLEFT", "l": 305.0648498535156, "r": 526.2252197265625, "t": 399.986083984375 }, "charspan": [ 0, 892 ], "page_no": 1 } ], "self_ref": "#/texts/10", "text": "The growth in datasets also brings significant challenges, including interface standardization, versioning, and documentation. A practitioner should be able to utilize N different datasets without requiring N different interfaces. In addition, N practitioners using the same dataset should know they have exactly the same version. Datasets have also grown larger, and ideally interfaces should not have to change due to this scale, whether one is using small-scale datasets like Climate Fever ( ∼ 1k data points), medium-scale Yahoo Answers ( ∼ 1M), or even all of PubMed ( ∼ 79B). Finally, datasets are being created with a variety of different procedures, from crowd-sourcing to scraping to synthetic generation, which need to be taken into account when evaluating which is most appropriate for a given purpose and ought to be immediately apparent to prospective users (Gebru et al., 2018)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets is a community library designed to address the challenges of dataset management and access, while supporting community culture and norms. The library targets the following goals:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 104.4854736328125, "coord_origin": "BOTTOMLEFT", "l": 305.25079345703125, "r": 526.2197875976562, "t": 156.06500244140625 }, "charspan": [ 0, 187 ], "page_no": 1 } ], "self_ref": "#/texts/11", "text": "Datasets is a community library designed to address the challenges of dataset management and access, while supporting community culture and norms. The library targets the following goals:" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Ease-of-use and Standardization: All datasets can be easily downloaded with one line of", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 67.83575439453125, "coord_origin": "BOTTOMLEFT", "l": 318.58123779296875, "r": 525.0678100585938, "t": 92.7308349609375 }, "charspan": [ 0, 89 ], "page_no": 1 } ], "self_ref": "#/texts/12", "text": "· Ease-of-use and Standardization: All datasets can be easily downloaded with one line of" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "code. Each dataset utilizes a standard tabular format, and is versioned and cited.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 744.1596069335938, "coord_origin": "BOTTOMLEFT", "l": 91.80062866210938, "r": 289.3184814453125, "t": 768.3035888671875 }, "charspan": [ 0, 82 ], "page_no": 2 } ], "self_ref": "#/texts/13", "text": "code. Each dataset utilizes a standard tabular format, and is versioned and cited." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Efficiency and Scale: Datasets are computationand memory-efficient by default and work seamlessly with tokenization and featurization. Massive datasets can even be streamed through the same interface.", "parent": { "$ref": "#/groups/1" }, "prov": [ { "bbox": { "b": 668.1622314453125, "coord_origin": "BOTTOMLEFT", "l": 83.02896118164062, "r": 290.94989013671875, "t": 733.0386352539062 }, "charspan": [ 0, 202 ], "page_no": 2 } ], "self_ref": "#/texts/14", "text": "· Efficiency and Scale: Datasets are computationand memory-efficient by default and work seamlessly with tokenization and featurization. Massive datasets can even be streamed through the same interface." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Community and Documentation: The project is community-built and has hundreds of contributors across languages. Each dataset is tagged and documented with a datasheet describing its usage, types, and construction.", "parent": { "$ref": "#/groups/1" }, "prov": [ { "bbox": { "b": 591.8584594726562, "coord_origin": "BOTTOMLEFT", "l": 83.0372085571289, "r": 290.9498596191406, "t": 657.7041015625 }, "charspan": [ 0, 214 ], "page_no": 2 } ], "self_ref": "#/texts/15", "text": "· Community and Documentation: The project is community-built and has hundreds of contributors across languages. Each dataset is tagged and documented with a datasheet describing its usage, types, and construction." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets is in continual development by the engineers at Hugging Face and is released under an Apache 2.0 license. 1 The library is available at https://github.com/huggingface/ datasets . Full documentation is available through the project website. 2", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 503.2791748046875, "coord_origin": "BOTTOMLEFT", "l": 69.77122497558594, "r": 290.9485778808594, "t": 582.1199951171875 }, "charspan": [ 0, 250 ], "page_no": 2 } ], "self_ref": "#/texts/16", "text": "Datasets is in continual development by the engineers at Hugging Face and is released under an Apache 2.0 license. 1 The library is available at https://github.com/huggingface/ datasets . Full documentation is available through the project website. 2" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "2 Related Work", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 480.234375, "coord_origin": "BOTTOMLEFT", "l": 70.13789367675781, "r": 160.0040740966797, "t": 491.6504211425781 }, "charspan": [ 0, 14 ], "page_no": 2 } ], "self_ref": "#/texts/17", "text": "2 Related Work" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "There is a long history of projects aiming to group, categorize, version, and distribute NLP datasets which we briefly survey. Most notably, the Linguistic Data Consortium (LDC) stores, serves, and manages a variety of datasets for language and speech. In addition to hosting and distributing corpus resources, the LDC supports significant annotation efforts. Other projects have aimed to collect related annotations together. Projects like OntoNotes have collected annotations across multiple tasks for a single corpus (Pradhan and Xue, 2009) whereas the Universal Dependency treebank (Nivre et al., 2016) collects similar annotations across languages. In machine translation, projects like OPUS catalog the translation resources for many different languages. These differ from Datasets which collects and provides access to datasets in a content-agnostic way.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 243.1146240234375, "coord_origin": "BOTTOMLEFT", "l": 69.6716537475586, "r": 291.0462341308594, "t": 470.4585266113281 }, "charspan": [ 0, 861 ], "page_no": 2 } ], "self_ref": "#/texts/18", "text": "There is a long history of projects aiming to group, categorize, version, and distribute NLP datasets which we briefly survey. Most notably, the Linguistic Data Consortium (LDC) stores, serves, and manages a variety of datasets for language and speech. In addition to hosting and distributing corpus resources, the LDC supports significant annotation efforts. Other projects have aimed to collect related annotations together. Projects like OntoNotes have collected annotations across multiple tasks for a single corpus (Pradhan and Xue, 2009) whereas the Universal Dependency treebank (Nivre et al., 2016) collects similar annotations across languages. In machine translation, projects like OPUS catalog the translation resources for many different languages. These differ from Datasets which collects and provides access to datasets in a content-agnostic way." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Other projects have aimed to make it easy to access core NLP datasets. The influential NLTK project (Bird, 2006) provided a data library that makes it easy to download and access core datasets. SpaCy also provides a similar loading interface (Honnibal and Montani, 2017). In recent years, concurrent with the move towards deep learning, there has been a growth in large freely available datasets often with less precise annotation standards. This has motivated cloud-based repositories", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 107.095458984375, "coord_origin": "BOTTOMLEFT", "l": 69.66796875, "r": 290.94921875, "t": 240.5751953125 }, "charspan": [ 0, 485 ], "page_no": 2 } ], "self_ref": "#/texts/19", "text": "Other projects have aimed to make it easy to access core NLP datasets. The influential NLTK project (Bird, 2006) provided a data library that makes it easy to download and access core datasets. SpaCy also provides a similar loading interface (Honnibal and Montani, 2017). In recent years, concurrent with the move towards deep learning, there has been a growth in large freely available datasets often with less precise annotation standards. This has motivated cloud-based repositories" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{1}$Datasets themselves may utilize different licenses which are documented in the library.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 79.78025817871094, "coord_origin": "BOTTOMLEFT", "l": 70.04058837890625, "r": 289.1304931640625, "t": 99.509033203125 }, "charspan": [ 0, 93 ], "page_no": 2 } ], "self_ref": "#/texts/20", "text": "$^{1}$Datasets themselves may utilize different licenses which are documented in the library." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{2}$https://huggingface.co/docs/datasets/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.10003662109375, "coord_origin": "BOTTOMLEFT", "l": 82.7406005859375, "r": 286.06005859375, "t": 78.0384521484375 }, "charspan": [ 0, 43 ], "page_no": 2 } ], "self_ref": "#/texts/21", "text": "$^{2}$https://huggingface.co/docs/datasets/" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "of datasets. Initiatives like TensorFlow-Datasets (2021) and TorchText (2021) have collected various datasets in a common cloud format. This project began as a fork of TensorFlow-Datasets, but has diverged significantly.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 703.0259399414062, "coord_origin": "BOTTOMLEFT", "l": 305.1621398925781, "r": 524.5120239257812, "t": 768.406005859375 }, "charspan": [ 0, 220 ], "page_no": 2 } ], "self_ref": "#/texts/22", "text": "of datasets. Initiatives like TensorFlow-Datasets (2021) and TorchText (2021) have collected various datasets in a common cloud format. This project began as a fork of TensorFlow-Datasets, but has diverged significantly." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets differs from these projects along several axes. The project is decoupled from any modeling framework and provides a general-purpose tabular API. It focuses on NLP specifically and provides specialized types and structures for language constructs. Finally, it prioritizes community management and documentation through the dataset hub and data cards, and aims to provide access to a long-tail of datasets for many tasks and languages.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 581.1060791015625, "coord_origin": "BOTTOMLEFT", "l": 305.11651611328125, "r": 526.318115234375, "t": 700.867431640625 }, "charspan": [ 0, 442 ], "page_no": 2 } ], "self_ref": "#/texts/23", "text": "Datasets differs from these projects along several axes. The project is decoupled from any modeling framework and provides a general-purpose tabular API. It focuses on NLP specifically and provides specialized types and structures for language constructs. Finally, it prioritizes community management and documentation through the dataset hub and data cards, and aims to provide access to a long-tail of datasets for many tasks and languages." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3 Library Tour and Design", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 557.0001831054688, "coord_origin": "BOTTOMLEFT", "l": 305.378173828125, "r": 452.1747741699219, "t": 568.9898681640625 }, "charspan": [ 0, 25 ], "page_no": 2 } ], "self_ref": "#/texts/24", "text": "3 Library Tour and Design" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We begin with a brief tour. Accessing a dataset is done simply by referring to it by a global identity.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 523.0078735351562, "coord_origin": "BOTTOMLEFT", "l": 304.8360900878906, "r": 524.406494140625, "t": 547.5392456054688 }, "charspan": [ 0, 103 ], "page_no": 2 } ], "self_ref": "#/texts/25", "text": "We begin with a brief tour. Accessing a dataset is done simply by referring to it by a global identity." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dataset = load_dataset(\"boolq\")", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 508.6389465332031, "coord_origin": "BOTTOMLEFT", "l": 306.1419982910156, "r": 472.9171142578125, "t": 515.5609741210938 }, "charspan": [ 0, 31 ], "page_no": 2 } ], "self_ref": "#/texts/26", "text": "dataset = load_dataset(\"boolq\")" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Each dataset has a features schema and metadata.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 488.1746520996094, "coord_origin": "BOTTOMLEFT", "l": 305.4506530761719, "r": 521.8257446289062, "t": 498.9390563964844 }, "charspan": [ 0, 48 ], "page_no": 2 } ], "self_ref": "#/texts/27", "text": "Each dataset has a features schema and metadata." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "print(dataset.features, dataset.info)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 473.3729553222656, "coord_origin": "BOTTOMLEFT", "l": 306.1419982910156, "r": 505.19610595703125, "t": 480.2950134277344 }, "charspan": [ 0, 37 ], "page_no": 2 } ], "self_ref": "#/texts/28", "text": "print(dataset.features, dataset.info)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Any slice of data points can be accessed directly without loading the full dataset into memory.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 439.0171813964844, "coord_origin": "BOTTOMLEFT", "l": 305.0535583496094, "r": 524.7894897460938, "t": 463.5924987792969 }, "charspan": [ 0, 95 ], "page_no": 2 } ], "self_ref": "#/texts/29", "text": "Any slice of data points can be accessed directly without loading the full dataset into memory." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dataset[\"train\"][start:end]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 424.5579528808594, "coord_origin": "BOTTOMLEFT", "l": 306.1419982910156, "r": 451.397705078125, "t": 431.4800109863281 }, "charspan": [ 0, 27 ], "page_no": 2 } ], "self_ref": "#/texts/30", "text": "dataset[\"train\"][start:end]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Processing can be applied to every data point in a batched and parallel fashion using standard libraries such as NumPy or Torch.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 376.7602844238281, "coord_origin": "BOTTOMLEFT", "l": 305.3818664550781, "r": 526.2173461914062, "t": 414.9749755859375 }, "charspan": [ 0, 128 ], "page_no": 2 } ], "self_ref": "#/texts/31", "text": "Processing can be applied to every data point in a batched and parallel fashion using standard libraries such as NumPy or Torch." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "# Torch function \"tokenize\" tokenized = dataset.map(tokenize, num_proc=32)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 342.2679443359375, "coord_origin": "BOTTOMLEFT", "l": 306.1419982910156, "r": 499.81707763671875, "t": 369.1159973144531 }, "charspan": [ 0, 74 ], "page_no": 2 } ], "self_ref": "#/texts/32", "text": "# Torch function \"tokenize\" tokenized = dataset.map(tokenize, num_proc=32)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets facilitates each of these four Stages with the following technical steps.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 307.79345703125, "coord_origin": "BOTTOMLEFT", "l": 304.8583679199219, "r": 524.4054565429688, "t": 332.2096862792969 }, "charspan": [ 0, 82 ], "page_no": 2 } ], "self_ref": "#/texts/33", "text": "Datasets facilitates each of these four Stages with the following technical steps." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "S1. Dataset Retrieval and Building Datasets does not host the underlying raw datasets, but accesses hosted data from the original authors in a distributed manner. 3 Each dataset has a community contributed builder module. The builder module has the responsibility of processing the raw data, e.g. text or CSV, into a common dataset interface representation.", "parent": { "$ref": "#/groups/2" }, "prov": [ { "bbox": { "b": 191.66107177734375, "coord_origin": "BOTTOMLEFT", "l": 305.11578369140625, "r": 526.2174682617188, "t": 298.44439697265625 }, "charspan": [ 0, 357 ], "page_no": 2 } ], "self_ref": "#/texts/34", "text": "S1. Dataset Retrieval and Building Datasets does not host the underlying raw datasets, but accesses hosted data from the original authors in a distributed manner. 3 Each dataset has a community contributed builder module. The builder module has the responsibility of processing the raw data, e.g. text or CSV, into a common dataset interface representation." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "S2. Data Point Representation Each built dataset is represented internally as a table with typed columns. The Dataset type system includes a variety of common and NLP-targeted types. In addition to atomic values (int's, float's, string's,", "parent": { "$ref": "#/groups/2" }, "prov": [ { "bbox": { "b": 117.4434814453125, "coord_origin": "BOTTOMLEFT", "l": 305.07818603515625, "r": 525.7722778320312, "t": 182.5545654296875 }, "charspan": [ 0, 238 ], "page_no": 2 } ], "self_ref": "#/texts/35", "text": "S2. Data Point Representation Each built dataset is represented internally as a table with typed columns. The Dataset type system includes a variety of common and NLP-targeted types. In addition to atomic values (int's, float's, string's," }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{3}$For datasets with intensive preprocessing, such as Wikipedia, a preprocessed version is hosted. Datasets removed by the author are not centrally cached and become unavailable.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.92925262451172, "coord_origin": "BOTTOMLEFT", "l": 304.9753112792969, "r": 525.8942260742188, "t": 108.48284912109375 }, "charspan": [ 0, 181 ], "page_no": 2 } ], "self_ref": "#/texts/36", "text": "$^{3}$For datasets with intensive preprocessing, such as Wikipedia, a preprocessed version is hosted. Datasets removed by the author are not centrally cached and become unavailable." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "binary blobs) and JSON-like dicts and lists, the library also includes named categorical class labels, sequences, paired translations, and higherdimension arrays for images, videos, or waveforms.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 716.488525390625, "coord_origin": "BOTTOMLEFT", "l": 69.87886047363281, "r": 291.0454406738281, "t": 768.2763671875 }, "charspan": [ 0, 195 ], "page_no": 3 } ], "self_ref": "#/texts/37", "text": "binary blobs) and JSON-like dicts and lists, the library also includes named categorical class labels, sequences, paired translations, and higherdimension arrays for images, videos, or waveforms." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "S3. In-Memory Access Datasets is built on top of Apache Arrow, a cross-language columnar data framework (Arrow, 2020). Arrow provides a local caching system allowing datasets to be backed by an on-disk cache, which is memory-mapped for fast lookup. This architecture allows for large datasets to be used on machines with relatively small device memory. Arrow also allows for copy-free hand-offs to standard machine learning tools such as NumPy, Pandas, Torch, and TensorFlow.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 573.056640625, "coord_origin": "BOTTOMLEFT", "l": 69.85542297363281, "r": 290.4949951171875, "t": 705.7728881835938 }, "charspan": [ 0, 475 ], "page_no": 3 } ], "self_ref": "#/texts/38", "text": "S3. In-Memory Access Datasets is built on top of Apache Arrow, a cross-language columnar data framework (Arrow, 2020). Arrow provides a local caching system allowing datasets to be backed by an on-disk cache, which is memory-mapped for fast lookup. This architecture allows for large datasets to be used on machines with relatively small device memory. Arrow also allows for copy-free hand-offs to standard machine learning tools such as NumPy, Pandas, Torch, and TensorFlow." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "S4. User Processing At download, the library provides access to the typed data with minimal preprocessing. It provides functions for dataset manipulation including sorting, shuffling, splitting, and filtering. For complex manipulations, it provides a powerful map function that supports arbitrary Python functions for creating new in-memory tables. For large datasets, map can be run in batched, multi-process mode to apply processing in parallel. Furthermore, data processed by the same function is automatically cached between sessions.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 415.2926940917969, "coord_origin": "BOTTOMLEFT", "l": 70.0746078491211, "r": 291.04058837890625, "t": 561.5501708984375 }, "charspan": [ 0, 538 ], "page_no": 3 } ], "self_ref": "#/texts/39", "text": "S4. User Processing At download, the library provides access to the typed data with minimal preprocessing. It provides functions for dataset manipulation including sorting, shuffling, splitting, and filtering. For complex manipulations, it provides a powerful map function that supports arbitrary Python functions for creating new in-memory tables. For large datasets, map can be run in batched, multi-process mode to apply processing in parallel. Furthermore, data processed by the same function is automatically cached between sessions." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Complete Flow Upon requesting a dataset, it is downloaded from the original host. This triggers dataset-specific builder code which converts the text into a typed tabular format matching the feature schema and caches the table. The user is given a memory-mapped typed table. To further process the data, e.g. tokenize, the user can run arbitrary vectorized code and cache the results.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 298.59765625, "coord_origin": "BOTTOMLEFT", "l": 69.85218811035156, "r": 289.51708984375, "t": 404.24188232421875 }, "charspan": [ 0, 384 ], "page_no": 3 } ], "self_ref": "#/texts/40", "text": "Complete Flow Upon requesting a dataset, it is downloaded from the original host. This triggers dataset-specific builder code which converts the text into a typed tabular format matching the feature schema and caches the table. The user is given a memory-mapped typed table. To further process the data, e.g. tokenize, the user can run arbitrary vectorized code and cache the results." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "4 Dataset Documentation and Search", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 273.5403747558594, "coord_origin": "BOTTOMLEFT", "l": 69.67772674560547, "r": 268.8919677734375, "t": 285.0137939453125 }, "charspan": [ 0, 34 ], "page_no": 3 } ], "self_ref": "#/texts/41", "text": "4 Dataset Documentation and Search" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets is backed by the Dataset Hub 4 that helps users navigate the growing number of available resources and draws inspiration from recent work calling for better documentation of ML datasets in general (Gebru et al., 2018) and NLP datasets in particular (Bender and Friedman, 2018).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 183.978759765625, "coord_origin": "BOTTOMLEFT", "l": 69.85274505615234, "r": 289.4089660644531, "t": 263.6667175292969 }, "charspan": [ 0, 286 ], "page_no": 3 } ], "self_ref": "#/texts/42", "text": "Datasets is backed by the Dataset Hub 4 that helps users navigate the growing number of available resources and draws inspiration from recent work calling for better documentation of ML datasets in general (Gebru et al., 2018) and NLP datasets in particular (Bender and Friedman, 2018)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets can be seen as a form of infrastructure (Hutchinson et al., 2021). NLP practitioners typically make use of them with a specific goal in mind, whether they are looking to answer a specified research question or developing a system for a particular practical application. To that end, they need to be able to not only easily identify which", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 87.94354248046875, "coord_origin": "BOTTOMLEFT", "l": 70.1225357055664, "r": 290.94854736328125, "t": 180.9757080078125 }, "charspan": [ 0, 346 ], "page_no": 3 } ], "self_ref": "#/texts/43", "text": "Datasets can be seen as a form of infrastructure (Hutchinson et al., 2021). NLP practitioners typically make use of them with a specific goal in mind, whether they are looking to answer a specified research question or developing a system for a particular practical application. To that end, they need to be able to not only easily identify which" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{4}$https://hf.co/datasets/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.3192138671875, "coord_origin": "BOTTOMLEFT", "l": 82.23333740234375, "r": 210.7423095703125, "t": 78.86895751953125 }, "charspan": [ 0, 29 ], "page_no": 3 } ], "self_ref": "#/texts/44", "text": "$^{4}$https://hf.co/datasets/" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 1: The data card for ELI5 (Fan et al., 2019).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 570.20166015625, "coord_origin": "BOTTOMLEFT", "l": 312.7835388183594, "r": 517.1790161132812, "t": 580.09375 }, "charspan": [ 0, 52 ], "page_no": 3 } ], "self_ref": "#/texts/45", "text": "Figure 1: The data card for ELI5 (Fan et al., 2019)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "dataset is most appropriate for the task at hand, but also to understand how various properties of that best candidate might help with, or, conversely, run contrary to their purpose.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 490.2104797363281, "coord_origin": "BOTTOMLEFT", "l": 305.2630920410156, "r": 524.4147338867188, "t": 542.11767578125 }, "charspan": [ 0, 182 ], "page_no": 3 } ], "self_ref": "#/texts/46", "text": "dataset is most appropriate for the task at hand, but also to understand how various properties of that best candidate might help with, or, conversely, run contrary to their purpose." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The Dataset Hub includes all of the datasets available in the library. It links each of them together though: a set of structured tags holding information about their languages, tasks supported, licenses, etc.; a data card based on a template 5 designed to combine relevant technical considerations and broader context information (McMillan-Major et al., 2021); and a list of models trained on the dataset. Both the tags and data card are filled manually by the contributor who introduces the dataset to the library. Figure 1 presents an example of the dataset page on the hub. 6 Together, these pages and the search interface help users navigate the available resources.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 298.962646484375, "coord_origin": "BOTTOMLEFT", "l": 305.1385192871094, "r": 526.2246704101562, "t": 485.8343505859375 }, "charspan": [ 0, 671 ], "page_no": 3 } ], "self_ref": "#/texts/47", "text": "The Dataset Hub includes all of the datasets available in the library. It links each of them together though: a set of structured tags holding information about their languages, tasks supported, licenses, etc.; a data card based on a template 5 designed to combine relevant technical considerations and broader context information (McMillan-Major et al., 2021); and a list of models trained on the dataset. Both the tags and data card are filled manually by the contributor who introduces the dataset to the library. Figure 1 presents an example of the dataset page on the hub. 6 Together, these pages and the search interface help users navigate the available resources." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Choosing a Dataset Given a use case, the structured tags provide a way to surface helpful datasets. For example, requesting all datasets that have the tags for Spanish language and the Question Answering task category returns 7 items at the time of writing. A user can then refine their choice by reading through the data cards, which contain sections describing the variety of language used, legal considerations including licensing and incidence of Personal Identifying Information, and paragraphs about known social biases resulting from the collection process that might lead a deployed model to cause disparate harms.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 107.54595947265625, "coord_origin": "BOTTOMLEFT", "l": 304.9571533203125, "r": 526.3215942382812, "t": 281.031494140625 }, "charspan": [ 0, 622 ], "page_no": 3 } ], "self_ref": "#/texts/48", "text": "Choosing a Dataset Given a use case, the structured tags provide a way to surface helpful datasets. For example, requesting all datasets that have the tags for Spanish language and the Question Answering task category returns 7 items at the time of writing. A user can then refine their choice by reading through the data cards, which contain sections describing the variety of language used, legal considerations including licensing and incidence of Personal Identifying Information, and paragraphs about known social biases resulting from the collection process that might lead a deployed model to cause disparate harms." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{5}$https://hf.co/datasets/card-guide", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 79.6876220703125, "coord_origin": "BOTTOMLEFT", "l": 317.4678955078125, "r": 500.3178405761719, "t": 89.109375 }, "charspan": [ 0, 39 ], "page_no": 3 } ], "self_ref": "#/texts/49", "text": "$^{5}$https://hf.co/datasets/card-guide" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{6}$https://hf.co/datasets/eli5", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.484619140625, "coord_origin": "BOTTOMLEFT", "l": 317.780029296875, "r": 467.53668212890625, "t": 77.7457275390625 }, "charspan": [ 0, 33 ], "page_no": 3 } ], "self_ref": "#/texts/50", "text": "$^{6}$https://hf.co/datasets/eli5" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Using a Dataset The data card also contains information to help users navigate all the choices, from hardware to modeling, that go into successfully training a system. These include the number of examples in each of the dataset splits, the size on disk of the data, meaningful differences between the training, validation, and test split, and free text descriptions of the various fields that make up each example to help decide what information to use as input or output of a prediction model.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 635.0433349609375, "coord_origin": "BOTTOMLEFT", "l": 69.86194610595703, "r": 290.9433288574219, "t": 768.262451171875 }, "charspan": [ 0, 494 ], "page_no": 4 } ], "self_ref": "#/texts/51", "text": "Using a Dataset The data card also contains information to help users navigate all the choices, from hardware to modeling, that go into successfully training a system. These include the number of examples in each of the dataset splits, the size on disk of the data, meaningful differences between the training, validation, and test split, and free text descriptions of the various fields that make up each example to help decide what information to use as input or output of a prediction model." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The Data Card as a Living Document A dataset's life continues beyond its initial release. As NLP practitioners interact with the dataset in various ways, they may surface annotation artifacts that affect the behavior of trained models in unexpected ways (Gururangan et al., 2018), 7 issues in the way the standard split was initially devised to test a model's ability to adapt to new settings (Krishna et al., 2021), or new understanding of the social biases exhibited therein (Hutchinson et al., 2020). The community-driven nature of Datasets and the versioning mechanisms provided by the GitHub backend provide an opportunity to keep the data cards up to date as information comes to light and to make gradual progress toward having as complete documentation as possible.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 410.5079040527344, "coord_origin": "BOTTOMLEFT", "l": 69.7180404663086, "r": 291.04150390625, "t": 625.0443115234375 }, "charspan": [ 0, 773 ], "page_no": 4 } ], "self_ref": "#/texts/52", "text": "The Data Card as a Living Document A dataset's life continues beyond its initial release. As NLP practitioners interact with the dataset in various ways, they may surface annotation artifacts that affect the behavior of trained models in unexpected ways (Gururangan et al., 2018), 7 issues in the way the standard split was initially devised to test a model's ability to adapt to new settings (Krishna et al., 2021), or new understanding of the social biases exhibited therein (Hutchinson et al., 2020). The community-driven nature of Datasets and the versioning mechanisms provided by the GitHub backend provide an opportunity to keep the data cards up to date as information comes to light and to make gradual progress toward having as complete documentation as possible." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5 Dataset Usage and Use-Cases", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 387.49713134765625, "coord_origin": "BOTTOMLEFT", "l": 69.898681640625, "r": 237.9153594970703, "t": 399.7407531738281 }, "charspan": [ 0, 29 ], "page_no": 4 } ], "self_ref": "#/texts/53", "text": "5 Dataset Usage and Use-Cases" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets is now being actively used for a variety of tasks. Figure 2 (left) shows statistics about library usage. We can see that the most commonly downloaded libraries are popular English benchmarks such as GLUE and SQuAD which are often used for teaching and examples. However there is a range of popular models for different tasks and languages.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 273.0156555175781, "coord_origin": "BOTTOMLEFT", "l": 69.80762481689453, "r": 290.9413757324219, "t": 378.4273986816406 }, "charspan": [ 0, 348 ], "page_no": 4 } ], "self_ref": "#/texts/54", "text": "Datasets is now being actively used for a variety of tasks. Figure 2 (left) shows statistics about library usage. We can see that the most commonly downloaded libraries are popular English benchmarks such as GLUE and SQuAD which are often used for teaching and examples. However there is a range of popular models for different tasks and languages." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Figure 2 (right) shows the wide coverage of the library in terms of task types, sizes, and languages, with currently 681 total datasets. During the development of the Datasets project, there was a public hackathon to have community members develop new Dataset builders and add them to the project. This event led 485 commits and 285 unique contributors to the library. Recent work has outlined the difficulty of finding data sources for lowerresourced languages through automatic filtering alone (Caswell et al., 2021). The breadth of languages spoken by participants in this event made it possible to more reliably bootstrap the library", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 95.61358642578125, "coord_origin": "BOTTOMLEFT", "l": 69.7536849975586, "r": 291.0414733886719, "t": 270.52972412109375 }, "charspan": [ 0, 637 ], "page_no": 4 } ], "self_ref": "#/texts/55", "text": "Figure 2 (right) shows the wide coverage of the library in terms of task types, sizes, and languages, with currently 681 total datasets. During the development of the Datasets project, there was a public hackathon to have community members develop new Dataset builders and add them to the project. This event led 485 commits and 285 unique contributors to the library. Recent work has outlined the difficulty of finding data sources for lowerresourced languages through automatic filtering alone (Caswell et al., 2021). The breadth of languages spoken by participants in this event made it possible to more reliably bootstrap the library" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "$^{7}$https://hf.co/datasets/snli# other-known-limitations", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.47509765625, "coord_origin": "BOTTOMLEFT", "l": 70.3431167602539, "r": 237.6415252685547, "t": 88.54827880859375 }, "charspan": [ 0, 58 ], "page_no": 4 } ], "self_ref": "#/texts/56", "text": "$^{7}$https://hf.co/datasets/snli# other-known-limitations" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "with datasets in a wide range of different languages. Finally while Datasets is designed for NLP, it is becoming used for multi-modal datasets. The library now includes types for continuous data, including multi-dimensional arrays for image and video data and an Audio type.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 689.3369750976562, "coord_origin": "BOTTOMLEFT", "l": 305.2904052734375, "r": 526.3133544921875, "t": 768.2969970703125 }, "charspan": [ 0, 274 ], "page_no": 4 } ], "self_ref": "#/texts/57", "text": "with datasets in a wide range of different languages. Finally while Datasets is designed for NLP, it is becoming used for multi-modal datasets. The library now includes types for continuous data, including multi-dimensional arrays for image and video data and an Audio type." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5.1 Case Studies: N -Dataset NLP", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 666.5779418945312, "coord_origin": "BOTTOMLEFT", "l": 305.29388427734375, "r": 470.2869567871094, "t": 676.8807983398438 }, "charspan": [ 0, 32 ], "page_no": 4 } ], "self_ref": "#/texts/58", "text": "5.1 Case Studies: N -Dataset NLP" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "A standardized library of datasets opens up new use-cases beyond making single datasets easy to download. We highlight three use-cases in which practitioners have employed the Datasets library.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 606.9752197265625, "coord_origin": "BOTTOMLEFT", "l": 304.85931396484375, "r": 524.8005981445312, "t": 658.7622680664062 }, "charspan": [ 0, 193 ], "page_no": 4 } ], "self_ref": "#/texts/59", "text": "A standardized library of datasets opens up new use-cases beyond making single datasets easy to download. We highlight three use-cases in which practitioners have employed the Datasets library." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Case Study 1: N -task Pretraining Benchmarks Benchmarking frameworks such as NLP Decathlon and GLUE have popularized the comparison of a single NLP model across a variety of tasks (McCann et al., 2018; Wang et al., 2018). Recently benchmarking frameworks like GPT-3's test suite framework (Brown et al., 2020) have expanded this benchmarking style even further, taking on dozens of different tasks. This research has increased interest in comparison of different datasets at scale.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 463.8128356933594, "coord_origin": "BOTTOMLEFT", "l": 305.1656799316406, "r": 526.21728515625, "t": 596.527099609375 }, "charspan": [ 0, 481 ], "page_no": 4 } ], "self_ref": "#/texts/60", "text": "Case Study 1: N -task Pretraining Benchmarks Benchmarking frameworks such as NLP Decathlon and GLUE have popularized the comparison of a single NLP model across a variety of tasks (McCann et al., 2018; Wang et al., 2018). Recently benchmarking frameworks like GPT-3's test suite framework (Brown et al., 2020) have expanded this benchmarking style even further, taking on dozens of different tasks. This research has increased interest in comparison of different datasets at scale." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets is designed to facilitate large-scale, N -task benchmarking beyond what might be possible for a single researcher to set up. For example, the Eleuther AI project aims to produce a massive scale open-source model. As part of this project they have released an LM Evaluation Harness 8 which includes nearly 100 different NLP tasks to test a large scale language model. This framework is built with the Datasets library as a method for retrieving and caching datasets.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 327.8837890625, "coord_origin": "BOTTOMLEFT", "l": 305.05072021484375, "r": 526.2202758789062, "t": 460.8152770996094 }, "charspan": [ 0, 474 ], "page_no": 4 } ], "self_ref": "#/texts/61", "text": "Datasets is designed to facilitate large-scale, N -task benchmarking beyond what might be possible for a single researcher to set up. For example, the Eleuther AI project aims to produce a massive scale open-source model. As part of this project they have released an LM Evaluation Harness 8 which includes nearly 100 different NLP tasks to test a large scale language model. This framework is built with the Datasets library as a method for retrieving and caching datasets." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Case Study 2: Reproducible Shared Tasks NLP has a tradition of shared tasks that become long-lived benchmark datasets. Tasks like CoNLL 2000 (Tjong Kim Sang and Buchholz, 2000) continue to be widely used more than 20 years after their release. Datasets provides a convenient, reproducible, and standardized method for hosting and maintaining shared tasks, particularly when they require multiple different datasets.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 197.15936279296875, "coord_origin": "BOTTOMLEFT", "l": 305.1524658203125, "r": 526.2235107421875, "t": 317.22650146484375 }, "charspan": [ 0, 415 ], "page_no": 4 } ], "self_ref": "#/texts/62", "text": "Case Study 2: Reproducible Shared Tasks NLP has a tradition of shared tasks that become long-lived benchmark datasets. Tasks like CoNLL 2000 (Tjong Kim Sang and Buchholz, 2000) continue to be widely used more than 20 years after their release. Datasets provides a convenient, reproducible, and standardized method for hosting and maintaining shared tasks, particularly when they require multiple different datasets." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets was used to support the first GEM (Generation, Evaluation, and Metrics) workshop (Gehrmann et al., 2021). This workshop ran a shared task comparing natural language generation (NLG) systems on 12 different tasks. The tasks included examples from twenty different languages and supervised datasets varying from size of 5k examples to 500k. Critically, the shared task had", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 87.542724609375, "coord_origin": "BOTTOMLEFT", "l": 305.1143493652344, "r": 526.2245483398438, "t": 194.46575927734375 }, "charspan": [ 0, 379 ], "page_no": 4 } ], "self_ref": "#/texts/63", "text": "Datasets was used to support the first GEM (Generation, Evaluation, and Metrics) workshop (Gehrmann et al., 2021). This workshop ran a shared task comparing natural language generation (NLG) systems on 12 different tasks. The tasks included examples from twenty different languages and supervised datasets varying from size of 5k examples to 500k. Critically, the shared task had" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{8}$https://github.com/EleutherAI/lm-evaluation-harness", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.5718994140625, "coord_origin": "BOTTOMLEFT", "l": 318.00872802734375, "r": 511.49896240234375, "t": 78.623291015625 }, "charspan": [ 0, 57 ], "page_no": 4 } ], "self_ref": "#/texts/64", "text": "$^{8}$https://github.com/EleutherAI/lm-evaluation-harness" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 2: Summary statistics from the datasets in the library. ( Left ) The relative download numbers of the most popular datasets in the library. ( Right ) Task properties. Each dataset may have multiple sub-tasks. Task Types are the types labeled in the library. Task Sizes are the number of data points in the table. Task Languages are the languages tagged in the library (many datasets include tasks in different languages).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 138.20355224609375, "coord_origin": "BOTTOMLEFT", "l": 69.93646240234375, "r": 524.4349365234375, "t": 184.1607666015625 }, "charspan": [ 0, 428 ], "page_no": 5 } ], "self_ref": "#/texts/65", "text": "Figure 2: Summary statistics from the datasets in the library. ( Left ) The relative download numbers of the most popular datasets in the library. ( Right ) Task properties. Each dataset may have multiple sub-tasks. Task Types are the types labeled in the library. Task Sizes are the number of data points in the table. Task Languages are the languages tagged in the library (many datasets include tasks in different languages)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "a large variety of different input formats including tables, articles, RDF triples, and meaning graphs. Datasets allows users to access all 12 datasets with a single line of code in their shared task description.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 716.5720825195312, "coord_origin": "BOTTOMLEFT", "l": 69.873779296875, "r": 291.04541015625, "t": 768.251953125 }, "charspan": [ 0, 212 ], "page_no": 6 } ], "self_ref": "#/texts/66", "text": "a large variety of different input formats including tables, articles, RDF triples, and meaning graphs. Datasets allows users to access all 12 datasets with a single line of code in their shared task description." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Case Study 3: Robustness Evaluation While NLP models have improved to the point that on-paper they compete with human performance, many research projects have demonstrated that these same models are fooled when given out-ofdomain examples (Koehn and Knowles, 2017), simple adversarial constructions (Belinkov and Bisk, 2018), or examples that spuriously match basic patterns (Poliak et al., 2018).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 586.0115966796875, "coord_origin": "BOTTOMLEFT", "l": 69.79917907714844, "r": 290.94921875, "t": 705.1578369140625 }, "charspan": [ 0, 397 ], "page_no": 6 } ], "self_ref": "#/texts/67", "text": "Case Study 3: Robustness Evaluation While NLP models have improved to the point that on-paper they compete with human performance, many research projects have demonstrated that these same models are fooled when given out-ofdomain examples (Koehn and Knowles, 2017), simple adversarial constructions (Belinkov and Bisk, 2018), or examples that spuriously match basic patterns (Poliak et al., 2018)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Datasets can be used to support better benchmarking of these issues. The Robustness Gym 9 proposes a systematic way to test an NLP system across many different proposed techniques, specifically subpopulations, transformations, evaluation sets, and adversarial attacks (Goel et al., 2021). Together, these provide a robustness report that is more specific than a single evaluation measure. While developed independently, the Robustness Gym is built on Datasets , and \"relies on a common data interface\" provided by the library.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 435.6293029785156, "coord_origin": "BOTTOMLEFT", "l": 69.90453338623047, "r": 291.0415344238281, "t": 582.69580078125 }, "charspan": [ 0, 526 ], "page_no": 6 } ], "self_ref": "#/texts/68", "text": "Datasets can be used to support better benchmarking of these issues. The Robustness Gym 9 proposes a systematic way to test an NLP system across many different proposed techniques, specifically subpopulations, transformations, evaluation sets, and adversarial attacks (Goel et al., 2021). Together, these provide a robustness report that is more specific than a single evaluation measure. While developed independently, the Robustness Gym is built on Datasets , and \"relies on a common data interface\" provided by the library." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "6 Additional Functionality and Uses", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 410.57080078125, "coord_origin": "BOTTOMLEFT", "l": 70.1801986694336, "r": 263.1773376464844, "t": 422.3277587890625 }, "charspan": [ 0, 35 ], "page_no": 6 } ], "self_ref": "#/texts/69", "text": "6 Additional Functionality and Uses" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Streaming Some datasets are extremely large and cannot even fit on disk. Datasets includes a streaming mode that buffers these datasets on the fly. This mode supports the core map primitive, which works on each data batch as it is streamed. Datasets streaming helped enable recent research into distributed training of a very large open NLP model (Diskin et al., 2021).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 293.9556579589844, "coord_origin": "BOTTOMLEFT", "l": 69.98976135253906, "r": 291.0379333496094, "t": 399.4517822265625 }, "charspan": [ 0, 369 ], "page_no": 6 } ], "self_ref": "#/texts/70", "text": "Streaming Some datasets are extremely large and cannot even fit on disk. Datasets includes a streaming mode that buffers these datasets on the fly. This mode supports the core map primitive, which works on each data batch as it is streamed. Datasets streaming helped enable recent research into distributed training of a very large open NLP model (Diskin et al., 2021)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Indexing Datasets includes tools for easily building and utilizing a search index over an arbitrary dataset. To construct the index the library can interface either with FAISS or ElasticSearch (Johnson et al., 2017; Elastic, 2021). This interface makes it easy to efficiently find nearest neighbors either with textual or vector queries. Indexing was used to host the open-source version of Retrieval-Augmented Generation (Lewis et al., 2020), a generation model backed by the ability to query knowledge from large-scale knowledge sources.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 143.97186279296875, "coord_origin": "BOTTOMLEFT", "l": 69.8917236328125, "r": 290.94927978515625, "t": 290.0091552734375 }, "charspan": [ 0, 539 ], "page_no": 6 } ], "self_ref": "#/texts/71", "text": "Indexing Datasets includes tools for easily building and utilizing a search index over an arbitrary dataset. To construct the index the library can interface either with FAISS or ElasticSearch (Johnson et al., 2017; Elastic, 2021). This interface makes it easy to efficiently find nearest neighbors either with textual or vector queries. Indexing was used to host the open-source version of Retrieval-Augmented Generation (Lewis et al., 2020), a generation model backed by the ability to query knowledge from large-scale knowledge sources." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Metrics Datasets includes an interface for standardizing metrics which can be documented, versioned and matched with datasets. This functionality is particularly useful for benchmark datasets", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 88.52691650390625, "coord_origin": "BOTTOMLEFT", "l": 70.15274810791016, "r": 290.9435729980469, "t": 141.10546875 }, "charspan": [ 0, 191 ], "page_no": 6 } ], "self_ref": "#/texts/72", "text": "Metrics Datasets includes an interface for standardizing metrics which can be documented, versioned and matched with datasets. This functionality is particularly useful for benchmark datasets" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{9}$https://robustnessgym.com/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.50616455078125, "coord_origin": "BOTTOMLEFT", "l": 82.50688934326172, "r": 185.8675994873047, "t": 78.37811279296875 }, "charspan": [ 0, 32 ], "page_no": 6 } ], "self_ref": "#/texts/73", "text": "$^{9}$https://robustnessgym.com/" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 3: Datasets viewer is an application that shows all rows for all datasets in the library. The interface allows users to change datasets, subsets, and splits, while seeing the dataset schema and metadata.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 534.7352294921875, "coord_origin": "BOTTOMLEFT", "l": 305.48681640625, "r": 526.0663452148438, "t": 580.53955078125 }, "charspan": [ 0, 210 ], "page_no": 6 } ], "self_ref": "#/texts/74", "text": "Figure 3: Datasets viewer is an application that shows all rows for all datasets in the library. The interface allows users to change datasets, subsets, and splits, while seeing the dataset schema and metadata." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "such as GLUE that include multiple tasks each with their own metric. Some metrics like BLEU and SQuAD are included directly in the library code, whereas others are linked to external packages. The library also allows for metrics to be applied in a distributed manner over the dataset.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 424.1026611328125, "coord_origin": "BOTTOMLEFT", "l": 305.33306884765625, "r": 525.7721557617188, "t": 502.002197265625 }, "charspan": [ 0, 284 ], "page_no": 6 } ], "self_ref": "#/texts/75", "text": "such as GLUE that include multiple tasks each with their own metric. Some metrics like BLEU and SQuAD are included directly in the library code, whereas others are linked to external packages. The library also allows for metrics to be applied in a distributed manner over the dataset." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Data Viewer A benefit of the standardized interface of the library is that it makes it trivial to build a cross-task dataset viewer. As an example, Hugging Face hosts a generic viewer for the entirety of datasets (Figure 3) $^{10}$. In this viewer, anyone on the web can open all almost 650 different datasets and view any example. Because the tables are typed, the viewer can easily show all component features, structured data, and multi-modal features.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 297.78765869140625, "coord_origin": "BOTTOMLEFT", "l": 305.13714599609375, "r": 526.2218627929688, "t": 416.89678955078125 }, "charspan": [ 0, 455 ], "page_no": 6 } ], "self_ref": "#/texts/76", "text": "Data Viewer A benefit of the standardized interface of the library is that it makes it trivial to build a cross-task dataset viewer. As an example, Hugging Face hosts a generic viewer for the entirety of datasets (Figure 3) $^{10}$. In this viewer, anyone on the web can open all almost 650 different datasets and view any example. Because the tables are typed, the viewer can easily show all component features, structured data, and multi-modal features." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "7 Conclusion", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 258.73638916015625, "coord_origin": "BOTTOMLEFT", "l": 305.25494384765625, "r": 381.2087097167969, "t": 270.46868896484375 }, "charspan": [ 0, 12 ], "page_no": 6 } ], "self_ref": "#/texts/77", "text": "7 Conclusion" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hugging Face Datasets is an open-source, community-driven library that standardizes the processing, distribution, and documentation of NLP datasets. The core library is designed to be easy to use, fast, and to use the same interface for datasets of varying size. At 650 datasets from over 250 contributors, it makes it easy to use standard datasets, has facilitated new use cases of cross-dataset NLP, and has advanced features for tasks like indexing and streaming large datasets.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 104.48419189453125, "coord_origin": "BOTTOMLEFT", "l": 305.00726318359375, "r": 526.2252807617188, "t": 237.4249267578125 }, "charspan": [ 0, 481 ], "page_no": 6 } ], "self_ref": "#/texts/78", "text": "Hugging Face Datasets is an open-source, community-driven library that standardizes the processing, distribution, and documentation of NLP datasets. The core library is designed to be easy to use, fast, and to use the same interface for datasets of varying size. At 650 datasets from over 250 contributors, it makes it easy to use standard datasets, has facilitated new use cases of cross-dataset NLP, and has advanced features for tasks like indexing and streaming large datasets." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{10}$https://huggingface.co/datasets/viewer/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.47125244140625, "coord_origin": "BOTTOMLEFT", "l": 315.6376647949219, "r": 462.1747741699219, "t": 78.62127685546875 }, "charspan": [ 0, 46 ], "page_no": 6 } ], "self_ref": "#/texts/79", "text": "$^{10}$https://huggingface.co/datasets/viewer/" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Acknowledgements", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 757.0633544921875, "coord_origin": "BOTTOMLEFT", "l": 69.75933074951172, "r": 169.69964599609375, "t": 768.999755859375 }, "charspan": [ 0, 16 ], "page_no": 7 } ], "self_ref": "#/texts/80", "text": "Acknowledgements" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "While organized by Hugging Face, Datasets is an open-source project driven by contributors. This work was only possible thanks to Charin Polpanumas, Cahya Wirawan, Jonatas Grosman, Thomas Hudson, Zaid Alyafeai, Rahul Chauhan, Vineeth S, Sandip, Yvonnegitau, Jared T Nielsen, Michal Jamry, Bharat Raghunathan, Ceyda Cinarel, David Adelani, Misbah Khan, Steven Liu, Vasudev Gupta, Matthew Bui, Abdul Rafay Khalid, Beth Tenorio, Eduardo Gonzalez Ponferrada, Harshal Mittal, Hugo Abonizio, Moussa Kamal Eddine, Stefan Schweter, Sumanth Doddapaneni, Yavuz Kömeço˘glu, Yusuke Mori, J-chim, Ontocord, Skyprince999, Vrindaprabhu, Jonathan Bragg, Philip May, Alexander Seifert, Ivanzidov, Jake Tae, Karim Foda, Mohamed Al Salti, Nick Doiron, Vinay, Czabo, Vblagoje, Nilansh Rajput, Abdulelah S. Al Mesfer, Akshay Bhardwaj, Amit Moryossef, Basava Sai Naga Viswa Chaitanya, Darek Kłeczek, Darshan Gandhi, Gustavo Aguilar, Hassan Ismail Fawaz, Jack Morris, Jamesg, Jonathan Chang, Karthik Bhaskar, Manan Dey, Maria Grandury, Michael A. Hedderich, Mounica Maddela, Nathan Cooper, Purvi M, Richard Wang, Song Feng, Sourab Mangrulkar, Tanmoy, Vijayasaradhi, Zacharysbrown, Chameleontk, Eusip, Jeromeku, Patpizio, Tuner007, Benjamin Van Der Burgh, Bharati Patidar, George Mihaila, Olivier, Tim Isbister, Alessandro Suglia, Ba¸sak Buluz Kömeço˘glu, Boris Dayma, Dariusz Kajtoch, Frankie Robertson, Jieyu, Mihaelagaman, Nikhil Bartwal, Param Bhavsar, Paullerner, Rachelker, Ricardo Rei, Sai, Sasha Rush, Suraj Parmar, Takuro Niitsuma, Taycir Yahmed, Tuan-phong Nguyen, Vladimir Gurevich, Alex, Calpt, Idoh, Justin-yan, Katnoria, Sileod, Avinash Swaminathan, Connor Mccarthy, Jungwhan Kim, Leo Zhao, Sanjay Kamath, (bill) Yuchen Lin, 2dot71mily, 8bitmp3, Abi Komma, Adam, Adeep Hande, Aditya Siddhant, Akash Kumar Gautam, Alaa Houimel, Alex Dong, Along, Anastasia Shimorina, Andre Barbosa, Anton Lozhkov, Antonio V Mendoza, Ashmeet Lamba, Ayushi Dalmia, Batjedi, Behçet ¸Sentürk, Bernardt Duvenhage, Binny Mathew, Birger Moëll, Blanc Ray, Bram Vanroy, Clément Rebuffel, Daniel Khashabi, David Fidalgo, David Wadden, Dhruv Kumar, Diwakar Mahajan, Elron Bandel, Emrah Budur, Fatima Haouari, Fraser Greenlee, Gergely Nemeth, Gowtham.r, Hemil Desai, Hiroki Nakayama, Ilham F Putra, Jannis Vam-", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.50962829589844, "coord_origin": "BOTTOMLEFT", "l": 69.67707061767578, "r": 291.0415344238281, "t": 743.224853515625 }, "charspan": [ 0, 2270 ], "page_no": 7 } ], "self_ref": "#/texts/81", "text": "While organized by Hugging Face, Datasets is an open-source project driven by contributors. This work was only possible thanks to Charin Polpanumas, Cahya Wirawan, Jonatas Grosman, Thomas Hudson, Zaid Alyafeai, Rahul Chauhan, Vineeth S, Sandip, Yvonnegitau, Jared T Nielsen, Michal Jamry, Bharat Raghunathan, Ceyda Cinarel, David Adelani, Misbah Khan, Steven Liu, Vasudev Gupta, Matthew Bui, Abdul Rafay Khalid, Beth Tenorio, Eduardo Gonzalez Ponferrada, Harshal Mittal, Hugo Abonizio, Moussa Kamal Eddine, Stefan Schweter, Sumanth Doddapaneni, Yavuz Kömeço˘glu, Yusuke Mori, J-chim, Ontocord, Skyprince999, Vrindaprabhu, Jonathan Bragg, Philip May, Alexander Seifert, Ivanzidov, Jake Tae, Karim Foda, Mohamed Al Salti, Nick Doiron, Vinay, Czabo, Vblagoje, Nilansh Rajput, Abdulelah S. Al Mesfer, Akshay Bhardwaj, Amit Moryossef, Basava Sai Naga Viswa Chaitanya, Darek Kłeczek, Darshan Gandhi, Gustavo Aguilar, Hassan Ismail Fawaz, Jack Morris, Jamesg, Jonathan Chang, Karthik Bhaskar, Manan Dey, Maria Grandury, Michael A. Hedderich, Mounica Maddela, Nathan Cooper, Purvi M, Richard Wang, Song Feng, Sourab Mangrulkar, Tanmoy, Vijayasaradhi, Zacharysbrown, Chameleontk, Eusip, Jeromeku, Patpizio, Tuner007, Benjamin Van Der Burgh, Bharati Patidar, George Mihaila, Olivier, Tim Isbister, Alessandro Suglia, Ba¸sak Buluz Kömeço˘glu, Boris Dayma, Dariusz Kajtoch, Frankie Robertson, Jieyu, Mihaelagaman, Nikhil Bartwal, Param Bhavsar, Paullerner, Rachelker, Ricardo Rei, Sai, Sasha Rush, Suraj Parmar, Takuro Niitsuma, Taycir Yahmed, Tuan-phong Nguyen, Vladimir Gurevich, Alex, Calpt, Idoh, Justin-yan, Katnoria, Sileod, Avinash Swaminathan, Connor Mccarthy, Jungwhan Kim, Leo Zhao, Sanjay Kamath, (bill) Yuchen Lin, 2dot71mily, 8bitmp3, Abi Komma, Adam, Adeep Hande, Aditya Siddhant, Akash Kumar Gautam, Alaa Houimel, Alex Dong, Along, Anastasia Shimorina, Andre Barbosa, Anton Lozhkov, Antonio V Mendoza, Ashmeet Lamba, Ayushi Dalmia, Batjedi, Behçet ¸Sentürk, Bernardt Duvenhage, Binny Mathew, Birger Moëll, Blanc Ray, Bram Vanroy, Clément Rebuffel, Daniel Khashabi, David Fidalgo, David Wadden, Dhruv Kumar, Diwakar Mahajan, Elron Bandel, Emrah Budur, Fatima Haouari, Fraser Greenlee, Gergely Nemeth, Gowtham.r, Hemil Desai, Hiroki Nakayama, Ilham F Putra, Jannis Vam-" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "e La Rosa, Javier-jimenez99, Jeff Hale, Jeff Yang, Joel Niklaus, John Miller, John Mollas, Joshua Adelman, Juan Julián Cea Morán, Kacper Łukawski, Koichi Miyamoto, Kushal Kedia, Laxya Agarwal, Leandro Von Werra, Loïc Estève, Luca Di Liello, Malik Altakrori, Manuel, Maramhasanain, Marcin Flis, Matteo Manica, Matthew Peters, Mehrdad Farahani, Merve Noyan, Mihai Ilie, Mitchell Gordon, Niccolò Campolungo, Nihal Harish, Noa Onoszko, Nora Belrose, Or Sharir, Oyvind Tafjord, Pewolf, Pariente Manuel, Pasquale Minervini, Pedro Ortiz Suárez, Pedro Lima, Pengcheng Yin, Petros Stavropoulos, Phil Wang, Philipp Christmann, Philipp Dufter, Philippe Laban, Pierre Colombo, Rahul Danu, Rabeeh Karimi Mahabadi, Remi Calizzano, Reshinth Adithyan, Rodion Martynov, Roman Tezikov, Sam Shleifer, Sava¸s Yıldırım, Sergey Mkrtchyan, Shubham Jain, Shubhambindal2017, Subhendu Ranjan Mishra, Taimur Ibrahim, Tanmay Thakur, Thomas Diggelmann, Théophile Blard, Tobias Slott, Tsvetomila Mihaylova, Vaibhav Adlakha, Vegar Andreas Bergum, Victor Velev, Vlad Lialin, Wilson Lee, Yang Wang, Yasir Abdurrohman, Yenting (Adam) Lin, Yixin Nie, Yoav Artzi, Yoni Gottesman, Yongrae Jo, Yuxiang Wu, Zhong Peixiang, Zihan Wang, Aditya2211, Alejandrocros, Andy Zou, Brainshawn, Cemilcengiz, Chutaklee, Gaurav Rai, Dhruvjoshi1998, Duttahritwik, Enod, Felixgwu, Ggdupont, Jerryishere, Jeswan, Lodgi, Lorinczb, Maxbartolo, Nathan Dahlberg, Neal, Ngdodd, Kristo, Onur Güngör, Ophelielacroix, Padipadou, and Phiwi.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 323.423828125, "coord_origin": "BOTTOMLEFT", "l": 304.719482421875, "r": 526.2252807617188, "t": 768.73876953125 }, "charspan": [ 0, 1476 ], "page_no": 7 } ], "self_ref": "#/texts/82", "text": "e La Rosa, Javier-jimenez99, Jeff Hale, Jeff Yang, Joel Niklaus, John Miller, John Mollas, Joshua Adelman, Juan Julián Cea Morán, Kacper Łukawski, Koichi Miyamoto, Kushal Kedia, Laxya Agarwal, Leandro Von Werra, Loïc Estève, Luca Di Liello, Malik Altakrori, Manuel, Maramhasanain, Marcin Flis, Matteo Manica, Matthew Peters, Mehrdad Farahani, Merve Noyan, Mihai Ilie, Mitchell Gordon, Niccolò Campolungo, Nihal Harish, Noa Onoszko, Nora Belrose, Or Sharir, Oyvind Tafjord, Pewolf, Pariente Manuel, Pasquale Minervini, Pedro Ortiz Suárez, Pedro Lima, Pengcheng Yin, Petros Stavropoulos, Phil Wang, Philipp Christmann, Philipp Dufter, Philippe Laban, Pierre Colombo, Rahul Danu, Rabeeh Karimi Mahabadi, Remi Calizzano, Reshinth Adithyan, Rodion Martynov, Roman Tezikov, Sam Shleifer, Sava¸s Yıldırım, Sergey Mkrtchyan, Shubham Jain, Shubhambindal2017, Subhendu Ranjan Mishra, Taimur Ibrahim, Tanmay Thakur, Thomas Diggelmann, Théophile Blard, Tobias Slott, Tsvetomila Mihaylova, Vaibhav Adlakha, Vegar Andreas Bergum, Victor Velev, Vlad Lialin, Wilson Lee, Yang Wang, Yasir Abdurrohman, Yenting (Adam) Lin, Yixin Nie, Yoav Artzi, Yoni Gottesman, Yongrae Jo, Yuxiang Wu, Zhong Peixiang, Zihan Wang, Aditya2211, Alejandrocros, Andy Zou, Brainshawn, Cemilcengiz, Chutaklee, Gaurav Rai, Dhruvjoshi1998, Duttahritwik, Enod, Felixgwu, Ggdupont, Jerryishere, Jeswan, Lodgi, Lorinczb, Maxbartolo, Nathan Dahlberg, Neal, Ngdodd, Kristo, Onur Güngör, Ophelielacroix, Padipadou, and Phiwi." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "References", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 287.6163635253906, "coord_origin": "BOTTOMLEFT", "l": 305.701904296875, "r": 361.6858825683594, "t": 299.317138671875 }, "charspan": [ 0, 10 ], "page_no": 7 } ], "self_ref": "#/texts/83", "text": "References" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Apache Arrow. 2020. Apache Arrow, a crosslanguage development platform for in-memory analytics. https://arrow.apache.org/ .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 247.68804931640625, "coord_origin": "BOTTOMLEFT", "l": 305.3643493652344, "r": 526.0664672851562, "t": 280.3084716796875 }, "charspan": [ 0, 123 ], "page_no": 7 } ], "self_ref": "#/texts/84", "text": "Apache Arrow. 2020. Apache Arrow, a crosslanguage development platform for in-memory analytics. https://arrow.apache.org/ ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Yonatan Belinkov and Yonatan Bisk. 2018. Synthetic and natural noise both break neural machine translation. In International Conference on Learning Representations .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 195.84405517578125, "coord_origin": "BOTTOMLEFT", "l": 305.5108642578125, "r": 526.0663452148438, "t": 238.22296142578125 }, "charspan": [ 0, 165 ], "page_no": 7 } ], "self_ref": "#/texts/85", "text": "Yonatan Belinkov and Yonatan Bisk. 2018. Synthetic and natural noise both break neural machine translation. In International Conference on Learning Representations ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Emily M. Bender and Batya Friedman. 2018. Data statements for natural language processing: Toward mitigating system bias and enabling better science. Transactions of the Association for Computational Linguistics , 6:587-604.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 132.279052734375, "coord_origin": "BOTTOMLEFT", "l": 305.6918640136719, "r": 526.156005859375, "t": 185.605712890625 }, "charspan": [ 0, 224 ], "page_no": 7 } ], "self_ref": "#/texts/86", "text": "Emily M. Bender and Batya Friedman. 2018. Data statements for natural language processing: Toward mitigating system bias and enabling better science. Transactions of the Association for Computational Linguistics , 6:587-604." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Steven Bird. 2006. NLTK: The Natural Language Toolkit. In Proceedings of the COLING/ACL 2006 Interactive Presentation Sessions , pages 69-72, Sydney, Australia. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 67.70794677734375, "coord_origin": "BOTTOMLEFT", "l": 305.5758972167969, "r": 526.0663452148438, "t": 122.39202880859375 }, "charspan": [ 0, 203 ], "page_no": 7 } ], "self_ref": "#/texts/87", "text": "Steven Bird. 2006. NLTK: The Natural Language Toolkit. In Proceedings of the COLING/ACL 2006 Interactive Presentation Sessions , pages 69-72, Sydney, Australia. Association for Computational Linguistics." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "P. Brown, J. Cocke, S. Della Pietra, V. Della Pietra, F. Jelinek, R. Mercer, and P. Roossin. 1988. A statistical approach to language translation. In Coling Budapest 1988 Volume 1: International Conference on Computational Linguistics .", "parent": { "$ref": "#/groups/3" }, "prov": [ { "bbox": { "b": 713.384765625, "coord_origin": "BOTTOMLEFT", "l": 70.07630157470703, "r": 290.7903137207031, "t": 767.6859130859375 }, "charspan": [ 0, 236 ], "page_no": 8 } ], "self_ref": "#/texts/88", "text": "P. Brown, J. Cocke, S. Della Pietra, V. Della Pietra, F. Jelinek, R. Mercer, and P. Roossin. 1988. A statistical approach to language translation. In Coling Budapest 1988 Volume 1: International Conference on Computational Linguistics ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. Language models are few-shot learners. In Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 541.1693725585938, "coord_origin": "BOTTOMLEFT", "l": 69.53489685058594, "r": 290.791015625, "t": 704.2418823242188 }, "charspan": [ 0, 694 ], "page_no": 8 } ], "self_ref": "#/texts/89", "text": "Tom B. Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-Voss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel M. Ziegler, Jeffrey Wu, Clemens Winter, Christopher Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott Gray, Benjamin Chess, Jack Clark, Christopher Berner, Sam McCandlish, Alec Radford, Ilya Sutskever, and Dario Amodei. 2020. Language models are few-shot learners. In Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Isaac Caswell, Julia Kreutzer, Lisa Wang, Ahsan Wahab, Daan van Esch, Nasanbayar Ulzii-Orshikh, Allahsera Tapo, Nishant Subramani, Artem Sokolov, Claytone Sikasote, Monang Setyawan, Supheakmungkol Sarin, Sokhar Samb, Benoît Sagot, Clara Rivera, Annette Rios, Isabel Papadimitriou, Salomey Osei, Pedro Javier Ortiz Suárez, Iroro Orife, Kelechi Ogueji, Rubungo Andre Niyongabo, Toan Q. Nguyen, Mathias Müller, André Müller, Shamsuddeen Hassan Muhammad, Nanda Muhammad, Ayanda Mnyakeni, Jamshidbek Mirzakhalov, Tapiwanashe Matangira, Colin Leong, Nze Lawson, Sneha Kudugunta, Yacine Jernite, Mathias Jenny, Orhan Firat, Bonaventure F. P. Dossou, Sakhile Dlamini, Nisansa de Silva, Sakine Çabuk Balli, Stella Biderman, Alessia Battisti, Ahmed Baruwa, Ankur Bapna, Pallavi Baljekar, Israel Abebe Azime, Ayodele Awokoya, Duygu Ataman, Orevaoghene Ahia, Oghenefego Ahia, Sweta Agrawal, and Mofetoluwa Adeyemi. 2021. Quality at a glance: An audit of web-crawled multilingual datasets. CoRR , abs/2103.12028.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 291.54864501953125, "coord_origin": "BOTTOMLEFT", "l": 70.52281188964844, "r": 290.7909851074219, "t": 531.0081176757812 }, "charspan": [ 0, 999 ], "page_no": 8 } ], "self_ref": "#/texts/90", "text": "Isaac Caswell, Julia Kreutzer, Lisa Wang, Ahsan Wahab, Daan van Esch, Nasanbayar Ulzii-Orshikh, Allahsera Tapo, Nishant Subramani, Artem Sokolov, Claytone Sikasote, Monang Setyawan, Supheakmungkol Sarin, Sokhar Samb, Benoît Sagot, Clara Rivera, Annette Rios, Isabel Papadimitriou, Salomey Osei, Pedro Javier Ortiz Suárez, Iroro Orife, Kelechi Ogueji, Rubungo Andre Niyongabo, Toan Q. Nguyen, Mathias Müller, André Müller, Shamsuddeen Hassan Muhammad, Nanda Muhammad, Ayanda Mnyakeni, Jamshidbek Mirzakhalov, Tapiwanashe Matangira, Colin Leong, Nze Lawson, Sneha Kudugunta, Yacine Jernite, Mathias Jenny, Orhan Firat, Bonaventure F. P. Dossou, Sakhile Dlamini, Nisansa de Silva, Sakine Çabuk Balli, Stella Biderman, Alessia Battisti, Ahmed Baruwa, Ankur Bapna, Pallavi Baljekar, Israel Abebe Azime, Ayodele Awokoya, Duygu Ataman, Orevaoghene Ahia, Oghenefego Ahia, Sweta Agrawal, and Mofetoluwa Adeyemi. 2021. Quality at a glance: An audit of web-crawled multilingual datasets. CoRR , abs/2103.12028." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) , pages 4171-4186, Minneapolis, Minnesota. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 184.24652099609375, "coord_origin": "BOTTOMLEFT", "l": 69.89824676513672, "r": 290.79034423828125, "t": 281.59637451171875 }, "charspan": [ 0, 419 ], "page_no": 8 } ], "self_ref": "#/texts/91", "text": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers) , pages 4171-4186, Minneapolis, Minnesota. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Thomas Diggelmann, Jordan Boyd-Graber, Jannis Bulian, Massimiliano Ciaramita, and Markus Leippold. 2020. Climate-fever: A dataset for verification of real-world climate claims.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 132.0336456298828, "coord_origin": "BOTTOMLEFT", "l": 69.69007873535156, "r": 290.8799743652344, "t": 174.5067138671875 }, "charspan": [ 0, 176 ], "page_no": 8 } ], "self_ref": "#/texts/92", "text": "Thomas Diggelmann, Jordan Boyd-Graber, Jannis Bulian, Massimiliano Ciaramita, and Markus Leippold. 2020. Climate-fever: A dataset for verification of real-world climate claims." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Michael Diskin, Alexey Bukhtiyarov, Max Ryabinin, Lucile Saulnier, Quentin Lhoest, Anton Sinitsin, Dmitry Popov, Dmitry Pyrkin, Maxim Kashirin, Alexander Borzunov, Albert Villanova del Moral, Denis Mazur, Ilia Kobelev, Yacine Jernite, Thomas", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.71463775634766, "coord_origin": "BOTTOMLEFT", "l": 70.1187973022461, "r": 290.3818664550781, "t": 122.144287109375 }, "charspan": [ 0, 241 ], "page_no": 8 } ], "self_ref": "#/texts/93", "text": "Michael Diskin, Alexey Bukhtiyarov, Max Ryabinin, Lucile Saulnier, Quentin Lhoest, Anton Sinitsin, Dmitry Popov, Dmitry Pyrkin, Maxim Kashirin, Alexander Borzunov, Albert Villanova del Moral, Denis Mazur, Ilia Kobelev, Yacine Jernite, Thomas" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Wolf, and Gennady Pekhimenko. 2021. Distributed deep learning in open collaborations.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 746.5289306640625, "coord_origin": "BOTTOMLEFT", "l": 315.5515441894531, "r": 524.412841796875, "t": 767.4352416992188 }, "charspan": [ 0, 85 ], "page_no": 8 } ], "self_ref": "#/texts/94", "text": "Wolf, and Gennady Pekhimenko. 2021. Distributed deep learning in open collaborations." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Elastic. 2021. Elastic Search. https://www. elastic.co/ .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 715.816650390625, "coord_origin": "BOTTOMLEFT", "l": 305.7205505371094, "r": 528.5936889648438, "t": 736.8988037109375 }, "charspan": [ 0, 57 ], "page_no": 8 } ], "self_ref": "#/texts/95", "text": "Elastic. 2021. Elastic Search. https://www. elastic.co/ ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Angela Fan, Yacine Jernite, Ethan Perez, David Grangier, Jason Weston, and Michael Auli. 2019. ELI5: Long form question answering. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics , pages 3558-3567, Florence, Italy. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 640.6856079101562, "coord_origin": "BOTTOMLEFT", "l": 305.2745056152344, "r": 526.0663452148438, "t": 705.3773193359375 }, "charspan": [ 0, 300 ], "page_no": 8 } ], "self_ref": "#/texts/96", "text": "Angela Fan, Yacine Jernite, Ethan Perez, David Grangier, Jason Weston, and Michael Auli. 2019. ELI5: Long form question answering. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics , pages 3558-3567, Florence, Italy. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jennifer Wortman Vaughan, Hanna M. Wallach, Hal Daumé III, and Kate Crawford. 2018. Datasheets for datasets. CoRR , abs/1803.09010.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 587.7877197265625, "coord_origin": "BOTTOMLEFT", "l": 305.26715087890625, "r": 526.1559448242188, "t": 630.0157470703125 }, "charspan": [ 0, 182 ], "page_no": 8 } ], "self_ref": "#/texts/97", "text": "Timnit Gebru, Jamie Morgenstern, Briana Vecchione, Jennifer Wortman Vaughan, Hanna M. Wallach, Hal Daumé III, and Kate Crawford. 2018. Datasheets for datasets. CoRR , abs/1803.09010." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Sebastian Gehrmann, Tosin P. Adewumi, Karmanya Aggarwal, Pawan Sasanka Ammanamanchi, Aremu Anuoluwapo, Antoine Bosselut, Khyathi Raghavi Chandu, Miruna Adriana Clinciu, Dipanjan Das, Kaustubh D. Dhole, Wanyu Du, Esin Durmus, Ondvrej Duvsek, Chris C. Emezue, Varun Gangal, Cristina Garbacea, T. Hashimoto, Yufang Hou, Yacine Jernite, Harsh Jhamtani, Yangfeng Ji, Shailza Jolly, Mihir Kale, Dhruv Kumar, Faisal Ladhak, Aman Madaan, Mounica Maddela, Khyati Mahajan, Saad Mahamood, Bodhisattwa Prasad Majumder, Pedro Henrique Martins, Angelina McMillan-Major, Simon Mille, Emiel van Miltenburg, Moin Nadeem, Shashi Narayan, V. Nikolaev, Rubungo Andre Niyongabo, Salomey Osei, Ankur P. Parikh, Laura Perez-Beltrachini, Niranjan Rao, Vikas Raunak, Juan Diego Rodríguez, Sashank Santhanam, João Sedoc, Thibault Sellam, Samira Shaikh, Anastasia Shimorina, Marco Antonio Sobrevilla Cabezudo, Hendrik Strobelt, Nishant Subramani, W. Xu, Diyi Yang, Akhila Yerukola, and Jiawei Zhou. 2021. The gem benchmark: Natural language generation, its evaluation and metrics. ArXiv , abs/2102.01672.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 326.51300048828125, "coord_origin": "BOTTOMLEFT", "l": 304.87799072265625, "r": 526.0662841796875, "t": 577.5675048828125 }, "charspan": [ 0, 1077 ], "page_no": 8 } ], "self_ref": "#/texts/98", "text": "Sebastian Gehrmann, Tosin P. Adewumi, Karmanya Aggarwal, Pawan Sasanka Ammanamanchi, Aremu Anuoluwapo, Antoine Bosselut, Khyathi Raghavi Chandu, Miruna Adriana Clinciu, Dipanjan Das, Kaustubh D. Dhole, Wanyu Du, Esin Durmus, Ondvrej Duvsek, Chris C. Emezue, Varun Gangal, Cristina Garbacea, T. Hashimoto, Yufang Hou, Yacine Jernite, Harsh Jhamtani, Yangfeng Ji, Shailza Jolly, Mihir Kale, Dhruv Kumar, Faisal Ladhak, Aman Madaan, Mounica Maddela, Khyati Mahajan, Saad Mahamood, Bodhisattwa Prasad Majumder, Pedro Henrique Martins, Angelina McMillan-Major, Simon Mille, Emiel van Miltenburg, Moin Nadeem, Shashi Narayan, V. Nikolaev, Rubungo Andre Niyongabo, Salomey Osei, Ankur P. Parikh, Laura Perez-Beltrachini, Niranjan Rao, Vikas Raunak, Juan Diego Rodríguez, Sashank Santhanam, João Sedoc, Thibault Sellam, Samira Shaikh, Anastasia Shimorina, Marco Antonio Sobrevilla Cabezudo, Hendrik Strobelt, Nishant Subramani, W. Xu, Diyi Yang, Akhila Yerukola, and Jiawei Zhou. 2021. The gem benchmark: Natural language generation, its evaluation and metrics. ArXiv , abs/2102.01672." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Karan Goel, Nazneen Fatema Rajani, Jesse Vig, Zachary Taschdjian, Mohit Bansal, and Christopher Ré. 2021. Robustness gym: Unifying the NLP evaluation landscape. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Demonstrations , pages 42-55, Online. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 229.56939697265625, "coord_origin": "BOTTOMLEFT", "l": 305.9731750488281, "r": 526.06640625, "t": 316.192626953125 }, "charspan": [ 0, 388 ], "page_no": 8 } ], "self_ref": "#/texts/99", "text": "Karan Goel, Nazneen Fatema Rajani, Jesse Vig, Zachary Taschdjian, Mohit Bansal, and Christopher Ré. 2021. Robustness gym: Unifying the NLP evaluation landscape. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Demonstrations , pages 42-55, Online. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Suchin Gururangan, Swabha Swayamdipta, Omer Levy, Roy Schwartz, Samuel Bowman, and Noah A. Smith. 2018. Annotation artifacts in natural language inference data. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers) , pages 107-112, New Orleans, Louisiana. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 121.3040771484375, "coord_origin": "BOTTOMLEFT", "l": 305.88330078125, "r": 526.1558837890625, "t": 218.8775634765625 }, "charspan": [ 0, 415 ], "page_no": 8 } ], "self_ref": "#/texts/100", "text": "Suchin Gururangan, Swabha Swayamdipta, Omer Levy, Roy Schwartz, Samuel Bowman, and Noah A. Smith. 2018. Annotation artifacts in natural language inference data. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers) , pages 107-112, New Orleans, Louisiana. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Matthew Honnibal and Ines Montani. 2017. spacy 2: Natural language understanding with bloom embeddings, convolutional neural networks and incremental parsing. To appear , 7(1):411-420.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 67.92724609375, "coord_origin": "BOTTOMLEFT", "l": 305.741943359375, "r": 526.0663452148438, "t": 111.82781982421875 }, "charspan": [ 0, 184 ], "page_no": 8 } ], "self_ref": "#/texts/101", "text": "Matthew Honnibal and Ines Montani. 2017. spacy 2: Natural language understanding with bloom embeddings, convolutional neural networks and incremental parsing. To appear , 7(1):411-420." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ben Hutchinson, Vinodkumar Prabhakaran, Emily Denton, Kellie Webster, Yu Zhong, and Stephen Denuyl. 2020. Social biases in NLP models as barriers for persons with disabilities. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, ACL 2020, Online, July 5-10, 2020 , pages 5491-5501. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 680.6940307617188, "coord_origin": "BOTTOMLEFT", "l": 70.29683685302734, "r": 290.8799743652344, "t": 767.5150146484375 }, "charspan": [ 0, 364 ], "page_no": 9 } ], "self_ref": "#/texts/102", "text": "Ben Hutchinson, Vinodkumar Prabhakaran, Emily Denton, Kellie Webster, Yu Zhong, and Stephen Denuyl. 2020. Social biases in NLP models as barriers for persons with disabilities. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, ACL 2020, Online, July 5-10, 2020 , pages 5491-5501. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ben Hutchinson, Andrew Smart, Alex Hanna, Emily Denton, Christina Greer, Oddur Kjartansson, Parker Barnes, and Margaret Mitchell. 2021. Towards accountability for machine learning datasets: Practices from software engineering and infrastructure. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency , FAccT '21, page 560-575, New York, NY, USA. Association for Computing Machinery.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 572.6550903320312, "coord_origin": "BOTTOMLEFT", "l": 70.16641235351562, "r": 290.790283203125, "t": 669.9539794921875 }, "charspan": [ 0, 416 ], "page_no": 9 } ], "self_ref": "#/texts/103", "text": "Ben Hutchinson, Andrew Smart, Alex Hanna, Emily Denton, Christina Greer, Oddur Kjartansson, Parker Barnes, and Margaret Mitchell. 2021. Towards accountability for machine learning datasets: Practices from software engineering and infrastructure. In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency , FAccT '21, page 560-575, New York, NY, USA. Association for Computing Machinery." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jeff Johnson, Matthijs Douze, and Hervé Jégou. 2017. Billion-scale similarity search with gpus. arXiv preprint arXiv:1702.08734 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 529.8502197265625, "coord_origin": "BOTTOMLEFT", "l": 69.71365356445312, "r": 290.88006591796875, "t": 562.265625 }, "charspan": [ 0, 129 ], "page_no": 9 } ], "self_ref": "#/texts/104", "text": "Jeff Johnson, Matthijs Douze, and Hervé Jégou. 2017. Billion-scale similarity search with gpus. arXiv preprint arXiv:1702.08734 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Philipp Koehn and Rebecca Knowles. 2017. Six challenges for neural machine translation. arXiv preprint arXiv:1706.03872 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 487.3158264160156, "coord_origin": "BOTTOMLEFT", "l": 70.27228546142578, "r": 289.3855895996094, "t": 519.516357421875 }, "charspan": [ 0, 121 ], "page_no": 9 } ], "self_ref": "#/texts/105", "text": "Philipp Koehn and Rebecca Knowles. 2017. Six challenges for neural machine translation. arXiv preprint arXiv:1706.03872 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kalpesh Krishna, Aurko Roy, and Mohit Iyyer. 2021. Hurdles to progress in long-form question answering. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2021, Online, June 6-11, 2021 , pages 4940-4957. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 390.2100830078125, "coord_origin": "BOTTOMLEFT", "l": 70.28902435302734, "r": 290.8800354003906, "t": 476.5027160644531 }, "charspan": [ 0, 352 ], "page_no": 9 } ], "self_ref": "#/texts/106", "text": "Kalpesh Krishna, Aurko Roy, and Mohit Iyyer. 2021. Hurdles to progress in long-form question answering. In Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2021, Online, June 6-11, 2021 , pages 4940-4957. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Patrick S. H. Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for knowledge-intensive NLP tasks. CoRR , abs/2005.11401.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 304.40155029296875, "coord_origin": "BOTTOMLEFT", "l": 70.37997436523438, "r": 290.7903747558594, "t": 380.19183349609375 }, "charspan": [ 0, 290 ], "page_no": 9 } ], "self_ref": "#/texts/107", "text": "Patrick S. H. Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, Sebastian Riedel, and Douwe Kiela. 2020. Retrieval-augmented generation for knowledge-intensive NLP tasks. CoRR , abs/2005.11401." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Mitchell P. Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. 1993. Building a large annotated corpus of English: The Penn Treebank. Computational Linguistics , 19(2):313-330.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 250.9585418701172, "coord_origin": "BOTTOMLEFT", "l": 70.00387573242188, "r": 290.7873840332031, "t": 293.6866455078125 }, "charspan": [ 0, 181 ], "page_no": 9 } ], "self_ref": "#/texts/108", "text": "Mitchell P. Marcus, Beatrice Santorini, and Mary Ann Marcinkiewicz. 1993. Building a large annotated corpus of English: The Penn Treebank. Computational Linguistics , 19(2):313-330." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Bryan McCann, N. Keskar, Caiming Xiong, and R. Socher. 2018. The natural language decathlon: Multitask learning as question answering. ArXiv , abs/1806.08730.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 197.51654052734375, "coord_origin": "BOTTOMLEFT", "l": 70.18072509765625, "r": 290.5213317871094, "t": 240.2064208984375 }, "charspan": [ 0, 158 ], "page_no": 9 } ], "self_ref": "#/texts/109", "text": "Bryan McCann, N. Keskar, Caiming Xiong, and R. Socher. 2018. The natural language decathlon: Multitask learning as question answering. ArXiv , abs/1806.08730." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Angelina McMillan-Major, Salomey Osey, Juan Diego Rodríguez, Pawan Sasanka Ammanamanchi, Sebastian Gehrmann, and Yacine Jernite. 2021. Reusable templates and guides for documenting datasets and models for natural language processing and generation - a case study of the huggingface and gem data and model cards.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 111.1975326538086, "coord_origin": "BOTTOMLEFT", "l": 70.45423126220703, "r": 290.79034423828125, "t": 186.564208984375 }, "charspan": [ 0, 311 ], "page_no": 9 } ], "self_ref": "#/texts/110", "text": "Angelina McMillan-Major, Salomey Osey, Juan Diego Rodríguez, Pawan Sasanka Ammanamanchi, Sebastian Gehrmann, and Yacine Jernite. 2021. Reusable templates and guides for documenting datasets and models for natural language processing and generation - a case study of the huggingface and gem data and model cards." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Joakim Nivre, Marie-Catherine de Marneffe, Filip Ginter, Yoav Goldberg, Jan Hajiˇc, Christopher D. Manning, Ryan McDonald, Slav Petrov, Sampo Pyysalo,", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.03143310546875, "coord_origin": "BOTTOMLEFT", "l": 69.90249633789062, "r": 290.7904052734375, "t": 100.2779541015625 }, "charspan": [ 0, 150 ], "page_no": 9 } ], "self_ref": "#/texts/111", "text": "Joakim Nivre, Marie-Catherine de Marneffe, Filip Ginter, Yoav Goldberg, Jan Hajiˇc, Christopher D. Manning, Ryan McDonald, Slav Petrov, Sampo Pyysalo," }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Natalia Silveira, Reut Tsarfaty, and Daniel Zeman. 2016. Universal Dependencies v1: A multilingual treebank collection. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16) , pages 1659-1666, Portorož, Slovenia. European Language Resources Association (ELRA).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 692.1596069335938, "coord_origin": "BOTTOMLEFT", "l": 315.92938232421875, "r": 526.156005859375, "t": 767.8448486328125 }, "charspan": [ 0, 306 ], "page_no": 9 } ], "self_ref": "#/texts/112", "text": "Natalia Silveira, Reut Tsarfaty, and Daniel Zeman. 2016. Universal Dependencies v1: A multilingual treebank collection. In Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16) , pages 1659-1666, Portorož, Slovenia. European Language Resources Association (ELRA)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Matthew E. Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) , pages 2227-2237, New Orleans, Louisiana. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 582.0691528320312, "coord_origin": "BOTTOMLEFT", "l": 305.30902099609375, "r": 526.0664672851562, "t": 680.3543701171875 }, "charspan": [ 0, 416 ], "page_no": 9 } ], "self_ref": "#/texts/113", "text": "Matthew E. Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers) , pages 2227-2237, New Orleans, Louisiana. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Adam Poliak, Jason Naradowsky, Aparajita Haldar, Rachel Rudinger, and Benjamin Van Durme. 2018. Hypothesis only baselines in natural language inference. In Proceedings of the Seventh Joint Conference on Lexical and Computational Semantics , pages 180-191, New Orleans, Louisiana. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 494.5164489746094, "coord_origin": "BOTTOMLEFT", "l": 305.6153259277344, "r": 526.1560668945312, "t": 570.9175415039062 }, "charspan": [ 0, 322 ], "page_no": 9 } ], "self_ref": "#/texts/114", "text": "Adam Poliak, Jason Naradowsky, Aparajita Haldar, Rachel Rudinger, and Benjamin Van Durme. 2018. Hypothesis only baselines in natural language inference. In Proceedings of the Seventh Joint Conference on Lexical and Computational Semantics , pages 180-191, New Orleans, Louisiana. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Sameer S. Pradhan and Nianwen Xue. 2009. OntoNotes: The 90% solution. In Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Tutorial Abstracts , pages 11-12, Boulder, Colorado. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 396.20257568359375, "coord_origin": "BOTTOMLEFT", "l": 305.4599609375, "r": 526.1561279296875, "t": 483.0164794921875 }, "charspan": [ 0, 337 ], "page_no": 9 } ], "self_ref": "#/texts/115", "text": "Sameer S. Pradhan and Nianwen Xue. 2009. OntoNotes: The 90% solution. In Proceedings of Human Language Technologies: The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Tutorial Abstracts , pages 11-12, Boulder, Colorado. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. 2020. Exploring the limits of transfer learning with a unified text-to-text transformer. J. Mach. Learn. Res. , 21:140:1-140:67.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 331.1229248046875, "coord_origin": "BOTTOMLEFT", "l": 305.0880432128906, "r": 526.06640625, "t": 384.6097106933594 }, "charspan": [ 0, 254 ], "page_no": 9 } ], "self_ref": "#/texts/116", "text": "Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. 2020. Exploring the limits of transfer learning with a unified text-to-text transformer. J. Mach. Learn. Res. , 21:140:1-140:67." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. 2016. SQuAD: 100,000+ questions for machine comprehension of text. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing , pages 2383-2392, Austin, Texas. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 254.09576416015625, "coord_origin": "BOTTOMLEFT", "l": 305.6480712890625, "r": 526.0665283203125, "t": 318.60626220703125 }, "charspan": [ 0, 300 ], "page_no": 9 } ], "self_ref": "#/texts/117", "text": "Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. 2016. SQuAD: 100,000+ questions for machine comprehension of text. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing , pages 2383-2392, Austin, Texas. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "TensorFlow-Datasets. 2021. TensorFlow Datasets, a collection of ready-to-use datasets. https:// www.tensorflow.org/datasets .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 210.87794494628906, "coord_origin": "BOTTOMLEFT", "l": 305.337158203125, "r": 525.6045532226562, "t": 241.9080810546875 }, "charspan": [ 0, 125 ], "page_no": 9 } ], "self_ref": "#/texts/118", "text": "TensorFlow-Datasets. 2021. TensorFlow Datasets, a collection of ready-to-use datasets. https:// www.tensorflow.org/datasets ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jörg Tiedemann and Lars Nygaard. 2004. The OPUS corpus - parallel and free: http://logos.uio. no/opus . In Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC'04) , Lisbon, Portugal. European Language Resources Association (ELRA).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 134.226806640625, "coord_origin": "BOTTOMLEFT", "l": 305.5015563964844, "r": 528.5936279296875, "t": 198.85699462890625 }, "charspan": [ 0, 272 ], "page_no": 9 } ], "self_ref": "#/texts/119", "text": "Jörg Tiedemann and Lars Nygaard. 2004. The OPUS corpus - parallel and free: http://logos.uio. no/opus . In Proceedings of the Fourth International Conference on Language Resources and Evaluation (LREC'04) , Lisbon, Portugal. European Language Resources Association (ELRA)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Erik F. Tjong Kim Sang and Sabine Buchholz. 2000. Introduction to the CoNLL-2000 shared task chunking. In Fourth Conference on Computational Natural Language Learning and the Second Learning Language in Logic Workshop .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.23223876953125, "coord_origin": "BOTTOMLEFT", "l": 305.7908935546875, "r": 526.15625, "t": 122.25616455078125 }, "charspan": [ 0, 219 ], "page_no": 9 } ], "self_ref": "#/texts/120", "text": "Erik F. Tjong Kim Sang and Sabine Buchholz. 2000. Introduction to the CoNLL-2000 shared task chunking. In Fourth Conference on Computational Natural Language Learning and the Second Learning Language in Logic Workshop ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "TorchText. 2021. TorchText. https://pytorch. org/text/stable/index.html .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 746.79248046875, "coord_origin": "BOTTOMLEFT", "l": 69.6768798828125, "r": 293.3179626464844, "t": 767.5140380859375 }, "charspan": [ 0, 73 ], "page_no": 10 } ], "self_ref": "#/texts/121", "text": "TorchText. 2021. TorchText. https://pytorch. org/text/stable/index.html ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Alex Wang, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel Bowman. 2018. GLUE: A multi-task benchmark and analysis platform for natural language understanding. In Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP , pages 353-355, Brussels, Belgium. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 649.77099609375, "coord_origin": "BOTTOMLEFT", "l": 69.84030151367188, "r": 290.8808898925781, "t": 737.2552490234375 }, "charspan": [ 0, 363 ], "page_no": 10 } ], "self_ref": "#/texts/122", "text": "Alex Wang, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel Bowman. 2018. GLUE: A multi-task benchmark and analysis platform for natural language understanding. In Proceedings of the 2018 EMNLP Workshop BlackboxNLP: Analyzing and Interpreting Neural Networks for NLP , pages 353-355, Brussels, Belgium. Association for Computational Linguistics." } ]

[ { "annotations": [], "captions": [], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 751.3611450195312, "coord_origin": "BOTTOMLEFT", "l": 78.02378845214844, "r": 104.0977554321289, "t": 774.8561401367188 }, "charspan": [ 0, 0 ], "page_no": 1 } ], "references": [], "self_ref": "#/pictures/0" }, { "annotations": [], "captions": [ { "$ref": "#/texts/45" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 588.7071533203125, "coord_origin": "BOTTOMLEFT", "l": 306.1444396972656, "r": 523.3623657226562, "t": 771.7030639648438 }, "charspan": [ 0, 52 ], "page_no": 3 } ], "references": [], "self_ref": "#/pictures/1" }, { "annotations": [], "captions": [ { "$ref": "#/texts/65" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 203.8900146484375, "coord_origin": "BOTTOMLEFT", "l": 84.23177337646484, "r": 513.5756225585938, "t": 706.010986328125 }, "charspan": [ 0, 428 ], "page_no": 5 } ], "references": [], "self_ref": "#/pictures/2" }, { "annotations": [], "captions": [ { "$ref": "#/texts/74" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 591.18798828125, "coord_origin": "BOTTOMLEFT", "l": 305.6175231933594, "r": 524.3847045898438, "t": 770.1265869140625 }, "charspan": [ 0, 210 ], "page_no": 6 } ], "references": [], "self_ref": "#/pictures/3" } ]

[]

[]

{ "1": { "page_no": 1, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "2": { "page_no": 2, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "3": { "page_no": 3, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "4": { "page_no": 4, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "5": { "page_no": 5, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "6": { "page_no": 6, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "7": { "page_no": 7, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "8": { "page_no": 8, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "9": { "page_no": 9, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "10": { "page_no": 10, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "11": null, "12": null, "13": null, "14": null, "15": null, "16": null, "17": null, "18": null, "19": null, "20": null, "21": null, "22": null, "23": null, "24": null, "25": null, "26": null, "27": null, "28": null, "29": null, "30": null, "31": null, "32": null, "33": null, "34": null, "35": null, "36": null }

{ "children": [], "label": "unspecified", "name": "_root_", "self_ref": "#/furniture" }

{ "children": [ { "$ref": "#/texts/0" }, { "$ref": "#/texts/1" }, { "$ref": "#/texts/2" }, { "$ref": "#/texts/3" }, { "$ref": "#/texts/4" }, { "$ref": "#/texts/5" }, { "$ref": "#/texts/6" }, { "$ref": "#/texts/7" }, { "$ref": "#/texts/8" }, { "$ref": "#/texts/9" }, { "$ref": "#/texts/10" }, { "$ref": "#/texts/11" }, { "$ref": "#/texts/12" }, { "$ref": "#/texts/13" }, { "$ref": "#/texts/14" }, { "$ref": "#/texts/15" }, { "$ref": "#/texts/16" }, { "$ref": "#/texts/17" }, { "$ref": "#/texts/18" }, { "$ref": "#/texts/19" }, { "$ref": "#/texts/20" }, { "$ref": "#/texts/21" }, { "$ref": "#/texts/22" }, { "$ref": "#/texts/23" }, { "$ref": "#/texts/24" }, { "$ref": "#/texts/25" }, { "$ref": "#/texts/26" }, { "$ref": "#/texts/27" }, { "$ref": "#/texts/28" }, { "$ref": "#/texts/29" }, { "$ref": "#/texts/30" }, { "$ref": "#/texts/31" }, { "$ref": "#/pictures/0" }, { "$ref": "#/texts/32" }, { "$ref": "#/texts/33" }, { "$ref": "#/texts/34" }, { "$ref": "#/texts/35" }, { "$ref": "#/texts/36" }, { "$ref": "#/texts/37" }, { "$ref": "#/texts/38" }, { "$ref": "#/texts/39" }, { "$ref": "#/pictures/1" }, { "$ref": "#/texts/40" }, { "$ref": "#/pictures/2" }, { "$ref": "#/texts/41" }, { "$ref": "#/texts/42" }, { "$ref": "#/texts/43" }, { "$ref": "#/texts/44" }, { "$ref": "#/texts/45" }, { "$ref": "#/texts/46" }, { "$ref": "#/texts/47" }, { "$ref": "#/texts/48" }, { "$ref": "#/texts/49" }, { "$ref": "#/texts/50" }, { "$ref": "#/texts/51" }, { "$ref": "#/texts/52" }, { "$ref": "#/texts/53" }, { "$ref": "#/texts/54" }, { "$ref": "#/texts/55" }, { "$ref": "#/texts/56" }, { "$ref": "#/texts/57" }, { "$ref": "#/texts/58" }, { "$ref": "#/groups/0" }, { "$ref": "#/texts/62" }, { "$ref": "#/texts/63" }, { "$ref": "#/texts/64" }, { "$ref": "#/texts/65" }, { "$ref": "#/texts/66" }, { "$ref": "#/texts/67" }, { "$ref": "#/texts/68" }, { "$ref": "#/texts/69" }, { "$ref": "#/tables/0" }, { "$ref": "#/texts/70" }, { "$ref": "#/texts/71" }, { "$ref": "#/texts/72" }, { "$ref": "#/texts/73" }, { "$ref": "#/texts/74" }, { "$ref": "#/texts/75" }, { "$ref": "#/texts/76" }, { "$ref": "#/texts/77" }, { "$ref": "#/texts/78" }, { "$ref": "#/texts/79" }, { "$ref": "#/texts/80" }, { "$ref": "#/texts/81" }, { "$ref": "#/texts/82" }, { "$ref": "#/texts/83" }, { "$ref": "#/texts/84" }, { "$ref": "#/texts/85" }, { "$ref": "#/texts/86" }, { "$ref": "#/texts/87" }, { "$ref": "#/texts/88" }, { "$ref": "#/texts/89" }, { "$ref": "#/texts/90" }, { "$ref": "#/texts/91" }, { "$ref": "#/texts/92" }, { "$ref": "#/texts/93" }, { "$ref": "#/texts/94" }, { "$ref": "#/texts/95" }, { "$ref": "#/texts/96" }, { "$ref": "#/texts/97" }, { "$ref": "#/texts/98" }, { "$ref": "#/tables/1" }, { "$ref": "#/texts/99" }, { "$ref": "#/texts/100" }, { "$ref": "#/texts/101" }, { "$ref": "#/texts/102" }, { "$ref": "#/texts/103" }, { "$ref": "#/texts/104" }, { "$ref": "#/texts/105" }, { "$ref": "#/texts/106" }, { "$ref": "#/texts/107" }, { "$ref": "#/texts/108" }, { "$ref": "#/texts/109" }, { "$ref": "#/texts/110" }, { "$ref": "#/texts/111" }, { "$ref": "#/tables/2" }, { "$ref": "#/texts/112" }, { "$ref": "#/texts/113" }, { "$ref": "#/texts/114" }, { "$ref": "#/texts/115" }, { "$ref": "#/texts/116" }, { "$ref": "#/texts/117" }, { "$ref": "#/texts/118" }, { "$ref": "#/texts/119" }, { "$ref": "#/texts/120" }, { "$ref": "#/tables/3" }, { "$ref": "#/texts/121" }, { "$ref": "#/texts/122" }, { "$ref": "#/texts/123" }, { "$ref": "#/texts/124" }, { "$ref": "#/texts/125" }, { "$ref": "#/texts/126" }, { "$ref": "#/texts/127" }, { "$ref": "#/texts/128" }, { "$ref": "#/texts/129" }, { "$ref": "#/texts/130" }, { "$ref": "#/groups/1" }, { "$ref": "#/texts/135" }, { "$ref": "#/tables/4" }, { "$ref": "#/texts/136" }, { "$ref": "#/tables/5" }, { "$ref": "#/texts/137" }, { "$ref": "#/texts/138" }, { "$ref": "#/texts/139" }, { "$ref": "#/pictures/3" }, { "$ref": "#/texts/140" }, { "$ref": "#/texts/141" }, { "$ref": "#/texts/142" }, { "$ref": "#/pictures/4" }, { "$ref": "#/texts/143" }, { "$ref": "#/texts/144" }, { "$ref": "#/texts/145" }, { "$ref": "#/pictures/5" }, { "$ref": "#/texts/146" } ], "label": "unspecified", "name": "_root_", "self_ref": "#/body" }

[ { "children": [ { "$ref": "#/texts/59" }, { "$ref": "#/texts/60" }, { "$ref": "#/texts/61" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/0" }, { "children": [ { "$ref": "#/texts/131" }, { "$ref": "#/texts/132" }, { "$ref": "#/texts/133" }, { "$ref": "#/texts/134" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/1" } ]

[ { "children": [], "enumerated": null, "label": "page_header", "level": null, "marker": null, "orig": "arXiv:1706.03762v7 [cs.CL] 2 Aug 2023", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 236.4552001953125, "coord_origin": "BOTTOMLEFT", "l": 17.49994468688965, "r": 36.339786529541016, "t": 578.0799560546875 }, "charspan": [ 0, 37 ], "page_no": 1 } ], "self_ref": "#/texts/0", "text": "arXiv:1706.03762v7 [cs.CL] 2 Aug 2023" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Provided proper attribution is provided, Google hereby grants permission to reproduce the tables and figures in this paper solely for use in journalistic or scholarly works.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 678.8435668945312, "coord_origin": "BOTTOMLEFT", "l": 123.57939910888672, "r": 487.89453125, "t": 719.0051879882812 }, "charspan": [ 0, 173 ], "page_no": 1 } ], "self_ref": "#/texts/1", "text": "Provided proper attribution is provided, Google hereby grants permission to reproduce the tables and figures in this paper solely for use in journalistic or scholarly works." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Attention Is All You Need", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 626.3589477539062, "coord_origin": "BOTTOMLEFT", "l": 210.61297607421875, "r": 399.89337158203125, "t": 642.747314453125 }, "charspan": [ 0, 25 ], "page_no": 1 } ], "self_ref": "#/texts/2", "text": "Attention Is All You Need" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ashish Vaswani ∗ Google Brain [email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 524.6845703125, "coord_origin": "BOTTOMLEFT", "l": 116.24988555908203, "r": 216.03900146484375, "t": 558.1503295898438 }, "charspan": [ 0, 49 ], "page_no": 1 } ], "self_ref": "#/texts/3", "text": "Ashish Vaswani ∗ Google Brain [email protected]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Noam Shazeer ∗ Google Brain [email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 524.8377075195312, "coord_origin": "BOTTOMLEFT", "l": 230.69200134277344, "r": 309.1325378417969, "t": 558.1503295898438 }, "charspan": [ 0, 43 ], "page_no": 1 } ], "self_ref": "#/texts/4", "text": "Noam Shazeer ∗ Google Brain [email protected]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Niki Parmar ∗ Google Research [email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 524.5406494140625, "coord_origin": "BOTTOMLEFT", "l": 323.77362060546875, "r": 407.5755310058594, "t": 558.1503295898438 }, "charspan": [ 0, 46 ], "page_no": 1 } ], "self_ref": "#/texts/5", "text": "Niki Parmar ∗ Google Research [email protected]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jakob Uszkoreit ∗ Google Research [email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 524.7680053710938, "coord_origin": "BOTTOMLEFT", "l": 422.11199951171875, "r": 497.2127685546875, "t": 558.1503295898438 }, "charspan": [ 0, 48 ], "page_no": 1 } ], "self_ref": "#/texts/6", "text": "Jakob Uszkoreit ∗ Google Research [email protected]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Llion Jones ∗ Google Research [email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 474.6589660644531, "coord_origin": "BOTTOMLEFT", "l": 125.96878814697266, "r": 210.71070861816406, "t": 508.1533203125 }, "charspan": [ 0, 46 ], "page_no": 1 } ], "self_ref": "#/texts/7", "text": "Llion Jones ∗ Google Research [email protected]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Aidan N. Gomez ∗ † University of Toronto [email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 475.27728271484375, "coord_origin": "BOTTOMLEFT", "l": 235.406982421875, "r": 339.9943542480469, "t": 508.1533203125 }, "charspan": [ 0, 61 ], "page_no": 1 } ], "self_ref": "#/texts/8", "text": "Aidan N. Gomez ∗ † University of Toronto [email protected]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Łukasz Kaiser ∗ Google Brain [email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 474.5601806640625, "coord_origin": "BOTTOMLEFT", "l": 364.1285705566406, "r": 485.4721984863281, "t": 508.1533203125 }, "charspan": [ 0, 52 ], "page_no": 1 } ], "self_ref": "#/texts/9", "text": "Łukasz Kaiser ∗ Google Brain [email protected]" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Illia Polosukhin ∗ ‡", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 447.2278137207031, "coord_origin": "BOTTOMLEFT", "l": 268.1126708984375, "r": 347.1256408691406, "t": 458.1563415527344 }, "charspan": [ 0, 20 ], "page_no": 1 } ], "self_ref": "#/texts/10", "text": "Illia Polosukhin ∗ ‡" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "[email protected]", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 436.0431213378906, "coord_origin": "BOTTOMLEFT", "l": 238.02200317382812, "r": 374.3076171875, "t": 445.7503967285156 }, "charspan": [ 0, 26 ], "page_no": 1 } ], "self_ref": "#/texts/11", "text": "[email protected]" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Abstract", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 394.7203674316406, "coord_origin": "BOTTOMLEFT", "l": 282.916015625, "r": 328.6545715332031, "t": 406.4098205566406 }, "charspan": [ 0, 8 ], "page_no": 1 } ], "self_ref": "#/texts/12", "text": "Abstract" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The dominant sequence transduction models are based on complex recurrent or convolutional neural networks that include an encoder and a decoder. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 Englishto-German translation task, improving over the existing best results, including ensembles, by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 215.08575439453125, "coord_origin": "BOTTOMLEFT", "l": 142.70025634765625, "r": 469.7870178222656, "t": 379.2144470214844 }, "charspan": [ 0, 1138 ], "page_no": 1 } ], "self_ref": "#/texts/13", "text": "The dominant sequence transduction models are based on complex recurrent or convolutional neural networks that include an encoder and a decoder. The best performing models also connect the encoder and decoder through an attention mechanism. We propose a new simple network architecture, the Transformer, based solely on attention mechanisms, dispensing with recurrence and convolutions entirely. Experiments on two machine translation tasks show these models to be superior in quality while being more parallelizable and requiring significantly less time to train. Our model achieves 28.4 BLEU on the WMT 2014 Englishto-German translation task, improving over the existing best results, including ensembles, by over 2 BLEU. On the WMT 2014 English-to-French translation task, our model establishes a new single-model state-of-the-art BLEU score of 41.8 after training for 3.5 days on eight GPUs, a small fraction of the training costs of the best models from the literature. We show that the Transformer generalizes well to other tasks by applying it successfully to English constituency parsing both with large and limited training data." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{∗}$Equal contribution. Listing order is random. Jakob proposed replacing RNNs with self-attention and started the effort to evaluate this idea. Ashish, with Illia, designed and implemented the first Transformer models and has been crucially involved in every aspect of this work. Noam proposed scaled dot-product attention, multi-head attention and the parameter-free position representation and became the other person involved in nearly every detail. Niki designed, implemented, tuned and evaluated countless model variants in our original codebase and tensor2tensor. Llion also experimented with novel model variants, was responsible for our initial codebase, and efficient inference and visualizations. Lukasz and Aidan spent countless long days designing various parts of and implementing tensor2tensor, replacing our earlier codebase, greatly improving results and massively accelerating our research.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 104.50726318359375, "coord_origin": "BOTTOMLEFT", "l": 107.05558013916016, "r": 504.4723205566406, "t": 193.1063232421875 }, "charspan": [ 0, 910 ], "page_no": 1 } ], "self_ref": "#/texts/14", "text": "$^{∗}$Equal contribution. Listing order is random. Jakob proposed replacing RNNs with self-attention and started the effort to evaluate this idea. Ashish, with Illia, designed and implemented the first Transformer models and has been crucially involved in every aspect of this work. Noam proposed scaled dot-product attention, multi-head attention and the parameter-free position representation and became the other person involved in nearly every detail. Niki designed, implemented, tuned and evaluated countless model variants in our original codebase and tensor2tensor. Llion also experimented with novel model variants, was responsible for our initial codebase, and efficient inference and visualizations. Lukasz and Aidan spent countless long days designing various parts of and implementing tensor2tensor, replacing our earlier codebase, greatly improving results and massively accelerating our research." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{†}$Work performed while at Google Brain.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 93.29962158203125, "coord_origin": "BOTTOMLEFT", "l": 120.0249252319336, "r": 267.4483947753906, "t": 102.759765625 }, "charspan": [ 0, 43 ], "page_no": 1 } ], "self_ref": "#/texts/15", "text": "$^{†}$Work performed while at Google Brain." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{‡}$Work performed while at Google Research.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 82.3494873046875, "coord_origin": "BOTTOMLEFT", "l": 119.56786346435547, "r": 280.28985595703125, "t": 92.13092041015625 }, "charspan": [ 0, 46 ], "page_no": 1 } ], "self_ref": "#/texts/16", "text": "$^{‡}$Work performed while at Google Research." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 49.81402587890625, "coord_origin": "BOTTOMLEFT", "l": 107.56991577148438, "r": 460.26934814453125, "t": 58.8609619140625 }, "charspan": [ 0, 90 ], "page_no": 1 } ], "self_ref": "#/texts/17", "text": "31st Conference on Neural Information Processing Systems (NIPS 2017), Long Beach, CA, USA." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "1 Introduction", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.538330078125, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 190.81365966796875, "t": 718.8709716796875 }, "charspan": [ 0, 14 ], "page_no": 2 } ], "self_ref": "#/texts/18", "text": "1 Introduction" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neural networks in particular, have been firmly established as state of the art approaches in sequence modeling and transduction problems such as language modeling and machine translation [35, 2, 5]. Numerous efforts have since continued to push the boundaries of recurrent language models and encoder-decoder architectures [38, 24, 15].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 639.5661010742188, "coord_origin": "BOTTOMLEFT", "l": 107.1638412475586, "r": 504.30316162109375, "t": 692.9063110351562 }, "charspan": [ 0, 416 ], "page_no": 2 } ], "self_ref": "#/texts/19", "text": "Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neural networks in particular, have been firmly established as state of the art approaches in sequence modeling and transduction problems such as language modeling and machine translation [35, 2, 5]. Numerous efforts have since continued to push the boundaries of recurrent language models and encoder-decoder architectures [38, 24, 15]." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Recurrent models typically factor computation along the symbol positions of the input and output sequences. Aligning the positions to steps in computation time, they generate a sequence of hidden states h$_{t}$ , as a function of the previous hidden state h$_{t}$$_{-}$$_{1}$ and the input for position t . This inherently sequential nature precludes parallelization within training examples, which becomes critical at longer sequence lengths, as memory constraints limit batching across examples. Recent work has achieved significant improvements in computational efficiency through factorization tricks [21] and conditional computation [32], while also improving model performance in case of the latter. The fundamental constraint of sequential computation, however, remains.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 546.5652465820312, "coord_origin": "BOTTOMLEFT", "l": 107.2396240234375, "r": 504.35089111328125, "t": 632.8406982421875 }, "charspan": [ 0, 777 ], "page_no": 2 } ], "self_ref": "#/texts/20", "text": "Recurrent models typically factor computation along the symbol positions of the input and output sequences. Aligning the positions to steps in computation time, they generate a sequence of hidden states h$_{t}$ , as a function of the previous hidden state h$_{t}$$_{-}$$_{1}$ and the input for position t . This inherently sequential nature precludes parallelization within training examples, which becomes critical at longer sequence lengths, as memory constraints limit batching across examples. Recent work has achieved significant improvements in computational efficiency through factorization tricks [21] and conditional computation [32], while also improving model performance in case of the latter. The fundamental constraint of sequential computation, however, remains." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Attention mechanisms have become an integral part of compelling sequence modeling and transduction models in various tasks, allowing modeling of dependencies without regard to their distance in the input or output sequences [2, 19]. In all but a few cases [27], however, such attention mechanisms are used in conjunction with a recurrent network.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 497.6980895996094, "coord_origin": "BOTTOMLEFT", "l": 107.21939849853516, "r": 505.6553649902344, "t": 540.0540771484375 }, "charspan": [ 0, 346 ], "page_no": 2 } ], "self_ref": "#/texts/21", "text": "Attention mechanisms have become an integral part of compelling sequence modeling and transduction models in various tasks, allowing modeling of dependencies without regard to their distance in the input or output sequences [2, 19]. In all but a few cases [27], however, such attention mechanisms are used in conjunction with a recurrent network." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In this work we propose the Transformer, a model architecture eschewing recurrence and instead relying entirely on an attention mechanism to draw global dependencies between input and output. The Transformer allows for significantly more parallelization and can reach a new state of the art in translation quality after being trained for as little as twelve hours on eight P100 GPUs.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 448.2679138183594, "coord_origin": "BOTTOMLEFT", "l": 106.93474578857422, "r": 505.73895263671875, "t": 490.8310241699219 }, "charspan": [ 0, 383 ], "page_no": 2 } ], "self_ref": "#/texts/22", "text": "In this work we propose the Transformer, a model architecture eschewing recurrence and instead relying entirely on an attention mechanism to draw global dependencies between input and output. The Transformer allows for significantly more parallelization and can reach a new state of the art in translation quality after being trained for as little as twelve hours on eight P100 GPUs." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "2 Background", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 417.2253723144531, "coord_origin": "BOTTOMLEFT", "l": 107.37066650390625, "r": 188.8291015625, "t": 428.97491455078125 }, "charspan": [ 0, 12 ], "page_no": 2 } ], "self_ref": "#/texts/23", "text": "2 Background" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The goal of reducing sequential computation also forms the foundation of the Extended Neural GPU [16], ByteNet [18] and ConvS2S [9], all of which use convolutional neural networks as basic building block, computing hidden representations in parallel for all input and output positions. In these models, the number of operations required to relate signals from two arbitrary input or output positions grows in the distance between positions, linearly for ConvS2S and logarithmically for ByteNet. This makes it more difficult to learn dependencies between distant positions [12]. In the Transformer this is reduced to a constant number of operations, albeit at the cost of reduced effective resolution due to averaging attention-weighted positions, an effect we counteract with Multi-Head Attention as described in section 3.2.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 305.6160888671875, "coord_origin": "BOTTOMLEFT", "l": 107.03768157958984, "r": 505.2413330078125, "t": 402.8783264160156 }, "charspan": [ 0, 825 ], "page_no": 2 } ], "self_ref": "#/texts/24", "text": "The goal of reducing sequential computation also forms the foundation of the Extended Neural GPU [16], ByteNet [18] and ConvS2S [9], all of which use convolutional neural networks as basic building block, computing hidden representations in parallel for all input and output positions. In these models, the number of operations required to relate signals from two arbitrary input or output positions grows in the distance between positions, linearly for ConvS2S and logarithmically for ByteNet. This makes it more difficult to learn dependencies between distant positions [12]. In the Transformer this is reduced to a constant number of operations, albeit at the cost of reduced effective resolution due to averaging attention-weighted positions, an effect we counteract with Multi-Head Attention as described in section 3.2." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Self-attention, sometimes called intra-attention is an attention mechanism relating different positions of a single sequence in order to compute a representation of the sequence. Self-attention has been used successfully in a variety of tasks including reading comprehension, abstractive summarization, textual entailment and learning task-independent sentence representations [4, 27, 28, 22].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 256.1927490234375, "coord_origin": "BOTTOMLEFT", "l": 107.24839782714844, "r": 505.2496032714844, "t": 298.7546691894531 }, "charspan": [ 0, 393 ], "page_no": 2 } ], "self_ref": "#/texts/25", "text": "Self-attention, sometimes called intra-attention is an attention mechanism relating different positions of a single sequence in order to compute a representation of the sequence. Self-attention has been used successfully in a variety of tasks including reading comprehension, abstractive summarization, textual entailment and learning task-independent sentence representations [4, 27, 28, 22]." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "End-to-end memory networks are based on a recurrent attention mechanism instead of sequencealigned recurrence and have been shown to perform well on simple-language question answering and language modeling tasks [34].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 218.2930908203125, "coord_origin": "BOTTOMLEFT", "l": 107.26404571533203, "r": 505.6534729003906, "t": 249.4962158203125 }, "charspan": [ 0, 217 ], "page_no": 2 } ], "self_ref": "#/texts/26", "text": "End-to-end memory networks are based on a recurrent attention mechanism instead of sequencealigned recurrence and have been shown to perform well on simple-language question answering and language modeling tasks [34]." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "To the best of our knowledge, however, the Transformer is the first transduction model relying entirely on self-attention to compute representations of its input and output without using sequencealigned RNNs or convolution. In the following sections, we will describe the Transformer, motivate self-attention and discuss its advantages over models such as [17, 18] and [9].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 169.1082763671875, "coord_origin": "BOTTOMLEFT", "l": 106.96324920654297, "r": 505.6572265625, "t": 211.41632080078125 }, "charspan": [ 0, 373 ], "page_no": 2 } ], "self_ref": "#/texts/27", "text": "To the best of our knowledge, however, the Transformer is the first transduction model relying entirely on self-attention to compute representations of its input and output without using sequencealigned RNNs or convolution. In the following sections, we will describe the Transformer, motivate self-attention and discuss its advantages over models such as [17, 18] and [9]." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3 Model Architecture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 137.8213653564453, "coord_origin": "BOTTOMLEFT", "l": 107.34628295898438, "r": 226.0934600830078, "t": 149.023681640625 }, "charspan": [ 0, 20 ], "page_no": 2 } ], "self_ref": "#/texts/28", "text": "3 Model Architecture" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Most competitive neural sequence transduction models have an encoder-decoder structure [5, 2, 35]. Here, the encoder maps an input sequence of symbol representations ( x$_{1}$, ..., x$_{n}$ ) to a sequence of continuous representations z = ( z$_{1}$, ..., z$_{n}$ ) . Given z , the decoder then generates an output sequence ( y$_{1}$, ..., y$_{m}$ ) of symbols one element at a time. At each step the model is auto-regressive [10], consuming the previously generated symbols as additional input when generating the next.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.68157958984375, "coord_origin": "BOTTOMLEFT", "l": 107.13375091552734, "r": 505.7431335449219, "t": 123.046875 }, "charspan": [ 0, 520 ], "page_no": 2 } ], "self_ref": "#/texts/29", "text": "Most competitive neural sequence transduction models have an encoder-decoder structure [5, 2, 35]. Here, the encoder maps an input sequence of symbol representations ( x$_{1}$, ..., x$_{n}$ ) to a sequence of continuous representations z = ( z$_{1}$, ..., z$_{n}$ ) . Given z , the decoder then generates an output sequence ( y$_{1}$, ..., y$_{m}$ ) of symbols one element at a time. At each step the model is auto-regressive [10], consuming the previously generated symbols as additional input when generating the next." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "2", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.89447021484375, "r": 308.49029541015625, "t": 49.83685302734375 }, "charspan": [ 0, 1 ], "page_no": 2 } ], "self_ref": "#/texts/30", "text": "2" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 1: The Transformer - model architecture.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 377.5470886230469, "coord_origin": "BOTTOMLEFT", "l": 209.3086700439453, "r": 401.9903564453125, "t": 387.3393249511719 }, "charspan": [ 0, 47 ], "page_no": 3 } ], "self_ref": "#/texts/31", "text": "Figure 1: The Transformer - model architecture." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The Transformer follows this overall architecture using stacked self-attention and point-wise, fully connected layers for both the encoder and decoder, shown in the left and right halves of Figure 1, respectively.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 323.6666564941406, "coord_origin": "BOTTOMLEFT", "l": 106.77662658691406, "r": 505.4501953125, "t": 354.7901306152344 }, "charspan": [ 0, 213 ], "page_no": 3 } ], "self_ref": "#/texts/32", "text": "The Transformer follows this overall architecture using stacked self-attention and point-wise, fully connected layers for both the encoder and decoder, shown in the left and right halves of Figure 1, respectively." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.1 Encoder and Decoder Stacks", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 299.2798156738281, "coord_origin": "BOTTOMLEFT", "l": 107.21050262451172, "r": 253.2667999267578, "t": 308.8091735839844 }, "charspan": [ 0, 30 ], "page_no": 3 } ], "self_ref": "#/texts/33", "text": "3.1 Encoder and Decoder Stacks" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Encoder: The encoder is composed of a stack of N = 6 identical layers. Each layer has two sub-layers. The first is a multi-head self-attention mechanism, and the second is a simple, positionwise fully connected feed-forward network. We employ a residual connection [11] around each of the two sub-layers, followed by layer normalization [1]. That is, the output of each sub-layer is LayerNorm( x + Sublayer( x )) , where Sublayer( x ) is the function implemented by the sub-layer itself. To facilitate these residual connections, all sub-layers in the model, as well as the embedding layers, produce outputs of dimension d$_{model}$ = 512 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 213.2666015625, "coord_origin": "BOTTOMLEFT", "l": 106.95099639892578, "r": 505.657470703125, "t": 288.8898620605469 }, "charspan": [ 0, 640 ], "page_no": 3 } ], "self_ref": "#/texts/34", "text": "Encoder: The encoder is composed of a stack of N = 6 identical layers. Each layer has two sub-layers. The first is a multi-head self-attention mechanism, and the second is a simple, positionwise fully connected feed-forward network. We employ a residual connection [11] around each of the two sub-layers, followed by layer normalization [1]. That is, the output of each sub-layer is LayerNorm( x + Sublayer( x )) , where Sublayer( x ) is the function implemented by the sub-layer itself. To facilitate these residual connections, all sub-layers in the model, as well as the embedding layers, produce outputs of dimension d$_{model}$ = 512 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Decoder: The decoder is also composed of a stack of N = 6 identical layers. In addition to the two sub-layers in each encoder layer, the decoder inserts a third sub-layer, which performs multi-head attention over the output of the encoder stack. Similar to the encoder, we employ residual connections around each of the sub-layers, followed by layer normalization. We also modify the self-attention sub-layer in the decoder stack to prevent positions from attending to subsequent positions. This masking, combined with fact that the output embeddings are offset by one position, ensures that the predictions for position i can depend only on the known outputs at positions less than i .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 124.503662109375, "coord_origin": "BOTTOMLEFT", "l": 107.15031433105469, "r": 504.1318359375, "t": 200.25701904296875 }, "charspan": [ 0, 686 ], "page_no": 3 } ], "self_ref": "#/texts/35", "text": "Decoder: The decoder is also composed of a stack of N = 6 identical layers. In addition to the two sub-layers in each encoder layer, the decoder inserts a third sub-layer, which performs multi-head attention over the output of the encoder stack. Similar to the encoder, we employ residual connections around each of the sub-layers, followed by layer normalization. We also modify the self-attention sub-layer in the decoder stack to prevent positions from attending to subsequent positions. This masking, combined with fact that the output embeddings are offset by one position, ensures that the predictions for position i can depend only on the known outputs at positions less than i ." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.2 Attention", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 100.8048095703125, "coord_origin": "BOTTOMLEFT", "l": 107.23251342773438, "r": 170.81419372558594, "t": 110.3282470703125 }, "charspan": [ 0, 13 ], "page_no": 3 } ], "self_ref": "#/texts/36", "text": "3.2 Attention" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "An attention function can be described as mapping a query and a set of key-value pairs to an output, where the query, keys, values, and output are all vectors. The output is computed as a weighted sum", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.423095703125, "coord_origin": "BOTTOMLEFT", "l": 106.86193084716797, "r": 505.24853515625, "t": 90.1895751953125 }, "charspan": [ 0, 200 ], "page_no": 3 } ], "self_ref": "#/texts/37", "text": "An attention function can be described as mapping a query and a set of key-value pairs to an output, where the query, keys, values, and output are all vectors. The output is computed as a weighted sum" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "3", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.75079345703125, "r": 308.49029541015625, "t": 49.50311279296875 }, "charspan": [ 0, 1 ], "page_no": 3 } ], "self_ref": "#/texts/38", "text": "3" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Scaled Dot-Product Attention", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 711.0930786132812, "coord_origin": "BOTTOMLEFT", "l": 147.7830047607422, "r": 266.2183837890625, "t": 719.9996337890625 }, "charspan": [ 0, 28 ], "page_no": 4 } ], "self_ref": "#/texts/39", "text": "Scaled Dot-Product Attention" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 2: (left) Scaled Dot-Product Attention. (right) Multi-Head Attention consists of several attention layers running in parallel.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 496.9200744628906, "coord_origin": "BOTTOMLEFT", "l": 107.47814178466797, "r": 503.9970703125, "t": 517.430419921875 }, "charspan": [ 0, 133 ], "page_no": 4 } ], "self_ref": "#/texts/40", "text": "Figure 2: (left) Scaled Dot-Product Attention. (right) Multi-Head Attention consists of several attention layers running in parallel." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "of the values, where the weight assigned to each value is computed by a compatibility function of the query with the corresponding key.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 453.8225402832031, "coord_origin": "BOTTOMLEFT", "l": 107.45587158203125, "r": 503.99493408203125, "t": 474.83038330078125 }, "charspan": [ 0, 135 ], "page_no": 4 } ], "self_ref": "#/texts/41", "text": "of the values, where the weight assigned to each value is computed by a compatibility function of the query with the corresponding key." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.2.1 Scaled Dot-Product Attention", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 431.2148132324219, "coord_origin": "BOTTOMLEFT", "l": 107.43103790283203, "r": 263.8847961425781, "t": 440.9424743652344 }, "charspan": [ 0, 34 ], "page_no": 4 } ], "self_ref": "#/texts/42", "text": "3.2.1 Scaled Dot-Product Attention" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We call our particular attention \"Scaled Dot-Product Attention\" (Figure 2). The input consists of queries and keys of dimension d$_{k}$ , and values of dimension d$_{v}$ . We compute the dot products of the query with all keys, divide each by √ d$_{k}$ , and apply a softmax function to obtain the weights on the values.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 379.7880859375, "coord_origin": "BOTTOMLEFT", "l": 107.02081298828125, "r": 504.2456970214844, "t": 422.29150390625 }, "charspan": [ 0, 320 ], "page_no": 4 } ], "self_ref": "#/texts/43", "text": "We call our particular attention \"Scaled Dot-Product Attention\" (Figure 2). The input consists of queries and keys of dimension d$_{k}$ , and values of dimension d$_{v}$ . We compute the dot products of the query with all keys, divide each by √ d$_{k}$ , and apply a softmax function to obtain the weights on the values." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In practice, we compute the attention function on a set of queries simultaneously, packed together into a matrix Q . The keys and values are also packed together into matrices K and V . We compute the matrix of outputs as:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 341.3092346191406, "coord_origin": "BOTTOMLEFT", "l": 107.43299865722656, "r": 504.16998291015625, "t": 372.7786560058594 }, "charspan": [ 0, 222 ], "page_no": 4 } ], "self_ref": "#/texts/44", "text": "In practice, we compute the attention function on a set of queries simultaneously, packed together into a matrix Q . The keys and values are also packed together into matrices K and V . We compute the matrix of outputs as:" }, { "children": [], "enumerated": null, "label": "formula", "level": null, "marker": null, "orig": "Attention( Q,K,V ) = softmax( QK T √ d$_{k}$ ) V (1)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 301.8502502441406, "coord_origin": "BOTTOMLEFT", "l": 219.61940002441406, "r": 504.6673889160156, "t": 326.9864807128906 }, "charspan": [ 0, 52 ], "page_no": 4 } ], "self_ref": "#/texts/45", "text": "Attention( Q,K,V ) = softmax( QK T √ d$_{k}$ ) V (1)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The two most commonly used attention functions are additive attention [2], and dot-product (multiplicative) attention. Dot-product attention is identical to our algorithm, except for the scaling factor of 1 √ $_{d$_{k}$}$. Additive attention computes the compatibility function using a feed-forward network with a single hidden layer. While the two are similar in theoretical complexity, dot-product attention is much faster and more space-efficient in practice, since it can be implemented using highly optimized matrix multiplication code.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 225.379638671875, "coord_origin": "BOTTOMLEFT", "l": 107.08892059326172, "r": 505.6528015136719, "t": 292.4774475097656 }, "charspan": [ 0, 541 ], "page_no": 4 } ], "self_ref": "#/texts/46", "text": "The two most commonly used attention functions are additive attention [2], and dot-product (multiplicative) attention. Dot-product attention is identical to our algorithm, except for the scaling factor of 1 √ $_{d$_{k}$}$. Additive attention computes the compatibility function using a feed-forward network with a single hidden layer. While the two are similar in theoretical complexity, dot-product attention is much faster and more space-efficient in practice, since it can be implemented using highly optimized matrix multiplication code." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "While for small values of d$_{k}$ the two mechanisms perform similarly, additive attention outperforms dot product attention without scaling for larger values of d$_{k}$ [3]. We suspect that for large values of d$_{k}$ , the dot products grow large in magnitude, pushing the softmax function into regions where it has extremely small gradients $^{4}$. To counteract this effect, we scale the dot products by 1 √ $_{d$_{k}$}$.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 175.2659912109375, "coord_origin": "BOTTOMLEFT", "l": 106.99662780761719, "r": 504.22869873046875, "t": 219.41986083984375 }, "charspan": [ 0, 425 ], "page_no": 4 } ], "self_ref": "#/texts/47", "text": "While for small values of d$_{k}$ the two mechanisms perform similarly, additive attention outperforms dot product attention without scaling for larger values of d$_{k}$ [3]. We suspect that for large values of d$_{k}$ , the dot products grow large in magnitude, pushing the softmax function into regions where it has extremely small gradients $^{4}$. To counteract this effect, we scale the dot products by 1 √ $_{d$_{k}$}$." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.2.2 Multi-Head Attention", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 151.49081420898438, "coord_origin": "BOTTOMLEFT", "l": 107.42676544189453, "r": 230.5897979736328, "t": 161.162353515625 }, "charspan": [ 0, 26 ], "page_no": 4 } ], "self_ref": "#/texts/48", "text": "3.2.2 Multi-Head Attention" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Instead of performing a single attention function with d$_{model}$ -dimensional keys, values and queries, we found it beneficial to linearly project the queries, keys and values h times with different, learned linear projections to d$_{k}$ , d$_{k}$ and d$_{v}$ dimensions, respectively. On each of these projected versions of queries, keys and values we then perform the attention function in parallel, yielding d$_{v}$ -dimensional", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 99.1368408203125, "coord_origin": "BOTTOMLEFT", "l": 107.28376770019531, "r": 505.2413330078125, "t": 142.05242919921875 }, "charspan": [ 0, 433 ], "page_no": 4 } ], "self_ref": "#/texts/49", "text": "Instead of performing a single attention function with d$_{model}$ -dimensional keys, values and queries, we found it beneficial to linearly project the queries, keys and values h times with different, learned linear projections to d$_{k}$ , d$_{k}$ and d$_{v}$ dimensions, respectively. On each of these projected versions of queries, keys and values we then perform the attention function in parallel, yielding d$_{v}$ -dimensional" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{4}$To illustrate why the dot products get large, assume that the components of q and k are independent random variables with mean 0 and variance 1 . Then their dot product, q · k = ∑ d$_{k}$ i $_{=1}$q$_{i}$ k$_{i}$ , has mean 0 and variance d$_{k}$ .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 68.10034942626953, "coord_origin": "BOTTOMLEFT", "l": 107.75934600830078, "r": 504.0268859863281, "t": 90.61468505859375 }, "charspan": [ 0, 254 ], "page_no": 4 } ], "self_ref": "#/texts/50", "text": "$^{4}$To illustrate why the dot products get large, assume that the components of q and k are independent random variables with mean 0 and variance 1 . Then their dot product, q · k = ∑ d$_{k}$ i $_{=1}$q$_{i}$ k$_{i}$ , has mean 0 and variance d$_{k}$ ." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "4", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.4430236816406, "r": 308.49029541015625, "t": 49.55816650390625 }, "charspan": [ 0, 1 ], "page_no": 4 } ], "self_ref": "#/texts/51", "text": "4" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "output values. These are concatenated and once again projected, resulting in the final values, as depicted in Figure 2.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 696.705078125, "coord_origin": "BOTTOMLEFT", "l": 107.42920684814453, "r": 503.9971923828125, "t": 717.6085205078125 }, "charspan": [ 0, 119 ], "page_no": 5 } ], "self_ref": "#/texts/52", "text": "output values. These are concatenated and once again projected, resulting in the final values, as depicted in Figure 2." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Multi-head attention allows the model to jointly attend to information from different representation subspaces at different positions. With a single attention head, averaging inhibits this.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 669.3574829101562, "coord_origin": "BOTTOMLEFT", "l": 107.50482177734375, "r": 503.9970397949219, "t": 690.1666259765625 }, "charspan": [ 0, 189 ], "page_no": 5 } ], "self_ref": "#/texts/53", "text": "Multi-head attention allows the model to jointly attend to information from different representation subspaces at different positions. With a single attention head, averaging inhibits this." }, { "children": [], "enumerated": null, "label": "formula", "level": null, "marker": null, "orig": "MultiHead( Q,K,V ) = Concat(head$_{1}$ ,..., head$_{h}$) W O where head$_{i}$ = Attention( QW Q i ,KW K i ,V W V i )", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 615.677001953125, "coord_origin": "BOTTOMLEFT", "l": 186.28192138671875, "r": 425.06048583984375, "t": 645.57080078125 }, "charspan": [ 0, 116 ], "page_no": 5 } ], "self_ref": "#/texts/54", "text": "MultiHead( Q,K,V ) = Concat(head$_{1}$ ,..., head$_{h}$) W O where head$_{i}$ = Attention( QW Q i ,KW K i ,V W V i )" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Where the projections are parameter matrices W Q i ∈$_{R}$ d$_{model}$ × $^{d$_{k}$}$, W K i ∈$_{R}$ d$_{model}$ × $^{d$_{k}$}$, W V i ∈$_{R}$ d$_{model}$ × d$_{v}$ and W O ∈$_{R}$ hd$_{v}$ × $^{d$_{model}$}$.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 564.30810546875, "coord_origin": "BOTTOMLEFT", "l": 106.81841278076172, "r": 502.8244934082031, "t": 588.59521484375 }, "charspan": [ 0, 209 ], "page_no": 5 } ], "self_ref": "#/texts/55", "text": "Where the projections are parameter matrices W Q i ∈$_{R}$ d$_{model}$ × $^{d$_{k}$}$, W K i ∈$_{R}$ d$_{model}$ × $^{d$_{k}$}$, W V i ∈$_{R}$ d$_{model}$ × d$_{v}$ and W O ∈$_{R}$ hd$_{v}$ × $^{d$_{model}$}$." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In this work we employ h = 8 parallel attention layers, or heads. For each of these we use d$_{k}$ = d$_{v}$ = d$_{model}$/h = 64 . Due to the reduced dimension of each head, the total computational cost is similar to that of single-head attention with full dimensionality.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 525.805908203125, "coord_origin": "BOTTOMLEFT", "l": 107.37702178955078, "r": 504.0018310546875, "t": 557.6996459960938 }, "charspan": [ 0, 273 ], "page_no": 5 } ], "self_ref": "#/texts/56", "text": "In this work we employ h = 8 parallel attention layers, or heads. For each of these we use d$_{k}$ = d$_{v}$ = d$_{model}$/h = 64 . Due to the reduced dimension of each head, the total computational cost is similar to that of single-head attention with full dimensionality." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.2.3 Applications of Attention in our Model", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 501.1728820800781, "coord_origin": "BOTTOMLEFT", "l": 107.39120483398438, "r": 302.8584899902344, "t": 511.77001953125 }, "charspan": [ 0, 44 ], "page_no": 5 } ], "self_ref": "#/texts/57", "text": "3.2.3 Applications of Attention in our Model" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The Transformer uses multi-head attention in three different ways:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 482.87506103515625, "coord_origin": "BOTTOMLEFT", "l": 106.91649627685547, "r": 372.6065368652344, "t": 492.7842102050781 }, "charspan": [ 0, 66 ], "page_no": 5 } ], "self_ref": "#/texts/58", "text": "The Transformer uses multi-head attention in three different ways:" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· In \"encoder-decoder attention\" layers, the queries come from the previous decoder layer, and the memory keys and values come from the output of the encoder. This allows every position in the decoder to attend over all positions in the input sequence. This mimics the typical encoder-decoder attention mechanisms in sequence-to-sequence models such as [38, 2, 9].", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 418.3940734863281, "coord_origin": "BOTTOMLEFT", "l": 134.55239868164062, "r": 505.2418212890625, "t": 471.5219421386719 }, "charspan": [ 0, 364 ], "page_no": 5 } ], "self_ref": "#/texts/59", "text": "· In \"encoder-decoder attention\" layers, the queries come from the previous decoder layer, and the memory keys and values come from the output of the encoder. This allows every position in the decoder to attend over all positions in the input sequence. This mimics the typical encoder-decoder attention mechanisms in sequence-to-sequence models such as [38, 2, 9]." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· The encoder contains self-attention layers. In a self-attention layer all of the keys, values and queries come from the same place, in this case, the output of the previous layer in the encoder. Each position in the encoder can attend to all positions in the previous layer of the encoder.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 369.8310852050781, "coord_origin": "BOTTOMLEFT", "l": 134.4431610107422, "r": 504.0082092285156, "t": 412.125244140625 }, "charspan": [ 0, 291 ], "page_no": 5 } ], "self_ref": "#/texts/60", "text": "· The encoder contains self-attention layers. In a self-attention layer all of the keys, values and queries come from the same place, in this case, the output of the previous layer in the encoder. Each position in the encoder can attend to all positions in the previous layer of the encoder." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "· Similarly, self-attention layers in the decoder allow each position in the decoder to attend to all positions in the decoder up to and including that position. We need to prevent leftward information flow in the decoder to preserve the auto-regressive property. We implement this inside of scaled dot-product attention by masking out (setting to -∞ ) all values in the input of the softmax which correspond to illegal connections. See Figure 2.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 309.932373046875, "coord_origin": "BOTTOMLEFT", "l": 134.66455078125, "r": 504.00299072265625, "t": 363.6373596191406 }, "charspan": [ 0, 446 ], "page_no": 5 } ], "self_ref": "#/texts/61", "text": "· Similarly, self-attention layers in the decoder allow each position in the decoder to attend to all positions in the decoder up to and including that position. We need to prevent leftward information flow in the decoder to preserve the auto-regressive property. We implement this inside of scaled dot-product attention by masking out (setting to -∞ ) all values in the input of the softmax which correspond to illegal connections. See Figure 2." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.3 Position-wise Feed-Forward Networks", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 284.9107971191406, "coord_origin": "BOTTOMLEFT", "l": 107.2830810546875, "r": 293.04522705078125, "t": 294.3228454589844 }, "charspan": [ 0, 39 ], "page_no": 5 } ], "self_ref": "#/texts/62", "text": "3.3 Position-wise Feed-Forward Networks" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In addition to attention sub-layers, each of the layers in our encoder and decoder contains a fully connected feed-forward network, which is applied to each position separately and identically. This consists of two linear transformations with a ReLU activation in between.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 242.62408447265625, "coord_origin": "BOTTOMLEFT", "l": 107.18673706054688, "r": 504.342529296875, "t": 274.35394287109375 }, "charspan": [ 0, 272 ], "page_no": 5 } ], "self_ref": "#/texts/63", "text": "In addition to attention sub-layers, each of the layers in our encoder and decoder contains a fully connected feed-forward network, which is applied to each position separately and identically. This consists of two linear transformations with a ReLU activation in between." }, { "children": [], "enumerated": null, "label": "formula", "level": null, "marker": null, "orig": "FFN( x ) = max(0 ,xW$_{1}$ + b$_{1}$ ) W$_{2}$ + b$_{2}$ (2)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 213.03070068359375, "coord_origin": "BOTTOMLEFT", "l": 226.48609924316406, "r": 504.9268798828125, "t": 224.16552734375 }, "charspan": [ 0, 60 ], "page_no": 5 } ], "self_ref": "#/texts/64", "text": "FFN( x ) = max(0 ,xW$_{1}$ + b$_{1}$ ) W$_{2}$ + b$_{2}$ (2)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "While the linear transformations are the same across different positions, they use different parameters from layer to layer. Another way of describing this is as two convolutions with kernel size 1. The dimensionality of input and output is d$_{model}$ = 512 , and the inner-layer has dimensionality d$_{ff}$ = 2048 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 158.10870361328125, "coord_origin": "BOTTOMLEFT", "l": 106.82971954345703, "r": 505.7450256347656, "t": 201.37506103515625 }, "charspan": [ 0, 317 ], "page_no": 5 } ], "self_ref": "#/texts/65", "text": "While the linear transformations are the same across different positions, they use different parameters from layer to layer. Another way of describing this is as two convolutions with kernel size 1. The dimensionality of input and output is d$_{model}$ = 512 , and the inner-layer has dimensionality d$_{ff}$ = 2048 ." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.4 Embeddings and Softmax", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 133.876953125, "coord_origin": "BOTTOMLEFT", "l": 107.43104553222656, "r": 240.0243377685547, "t": 143.5550537109375 }, "charspan": [ 0, 26 ], "page_no": 5 } ], "self_ref": "#/texts/66", "text": "3.4 Embeddings and Softmax" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Similarly to other sequence transduction models, we use learned embeddings to convert the input tokens and output tokens to vectors of dimension d$_{model}$ . We also use the usual learned linear transformation and softmax function to convert the decoder output to predicted next-token probabilities. In our model, we share the same weight matrix between the two embedding layers and the pre-softmax linear transformation, similar to [30]. In the embedding layers, we multiply those weights by √ d$_{model}$ .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 107.29226684570312, "r": 505.74383544921875, "t": 122.8626708984375 }, "charspan": [ 0, 509 ], "page_no": 5 } ], "self_ref": "#/texts/67", "text": "Similarly to other sequence transduction models, we use learned embeddings to convert the input tokens and output tokens to vectors of dimension d$_{model}$ . We also use the usual learned linear transformation and softmax function to convert the decoder output to predicted next-token probabilities. In our model, we share the same weight matrix between the two embedding layers and the pre-softmax linear transformation, similar to [30]. In the embedding layers, we multiply those weights by √ d$_{model}$ ." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "5", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.7465515136719, "r": 308.49029541015625, "t": 49.41009521484375 }, "charspan": [ 0, 1 ], "page_no": 5 } ], "self_ref": "#/texts/68", "text": "5" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Table 1: Maximum path lengths, per-layer complexity and minimum number of sequential operations for different layer types. n is the sequence length, d is the representation dimension, k is the kernel size of convolutions and r the size of the neighborhood in restricted self-attention.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 689.2750854492188, "coord_origin": "BOTTOMLEFT", "l": 107.1600112915039, "r": 504.0035400390625, "t": 720.675048828125 }, "charspan": [ 0, 285 ], "page_no": 6 } ], "self_ref": "#/texts/69", "text": "Table 1: Maximum path lengths, per-layer complexity and minimum number of sequential operations for different layer types. n is the sequence length, d is the representation dimension, k is the kernel size of convolutions and r the size of the neighborhood in restricted self-attention." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3.5 Positional Encoding", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 568.049072265625, "coord_origin": "BOTTOMLEFT", "l": 107.33191680908203, "r": 215.31039428710938, "t": 578.3384399414062 }, "charspan": [ 0, 23 ], "page_no": 6 } ], "self_ref": "#/texts/70", "text": "3.5 Positional Encoding" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Since our model contains no recurrence and no convolution, in order for the model to make use of the order of the sequence, we must inject some information about the relative or absolute position of the tokens in the sequence. To this end, we add \"positional encodings\" to the input embeddings at the bottoms of the encoder and decoder stacks. The positional encodings have the same dimension d$_{model}$ as the embeddings, so that the two can be summed. There are many choices of positional encodings, learned and fixed [9].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 493.8910827636719, "coord_origin": "BOTTOMLEFT", "l": 107.30216979980469, "r": 505.2479248046875, "t": 558.2435302734375 }, "charspan": [ 0, 525 ], "page_no": 6 } ], "self_ref": "#/texts/71", "text": "Since our model contains no recurrence and no convolution, in order for the model to make use of the order of the sequence, we must inject some information about the relative or absolute position of the tokens in the sequence. To this end, we add \"positional encodings\" to the input embeddings at the bottoms of the encoder and decoder stacks. The positional encodings have the same dimension d$_{model}$ as the embeddings, so that the two can be summed. There are many choices of positional encodings, learned and fixed [9]." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In this work, we use sine and cosine functions of different frequencies:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 476.7276611328125, "coord_origin": "BOTTOMLEFT", "l": 107.71692657470703, "r": 389.86187744140625, "t": 486.83758544921875 }, "charspan": [ 0, 72 ], "page_no": 6 } ], "self_ref": "#/texts/72", "text": "In this work, we use sine and cosine functions of different frequencies:" }, { "children": [], "enumerated": null, "label": "formula", "level": null, "marker": null, "orig": "PE$_{(}$$_{pos,}$$_{2}$$_{i}$$_{)}$ = sin ( pos/ 10000 2 $^{i/d$_{model}$}$) PE$_{(}$$_{pos,}$$_{2}$$_{i}$$_{+1)}$ = cos ( pos/ 10000 2 $^{i/d$_{model}$}$)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 425.7605285644531, "coord_origin": "BOTTOMLEFT", "l": 225.2696533203125, "r": 386.30645751953125, "t": 455.72772216796875 }, "charspan": [ 0, 155 ], "page_no": 6 } ], "self_ref": "#/texts/73", "text": "PE$_{(}$$_{pos,}$$_{2}$$_{i}$$_{)}$ = sin ( pos/ 10000 2 $^{i/d$_{model}$}$) PE$_{(}$$_{pos,}$$_{2}$$_{i}$$_{+1)}$ = cos ( pos/ 10000 2 $^{i/d$_{model}$}$)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "where pos is the position and i is the dimension. That is, each dimension of the positional encoding corresponds to a sinusoid. The wavelengths form a geometric progression from 2 π to 10000 · 2 π . We chose this function because we hypothesized it would allow the model to easily learn to attend by relative positions, since for any fixed offset k , PE$_{pos}$$_{+}$$_{k}$ can be represented as a linear function of PE$_{pos}$ .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 361.13507080078125, "coord_origin": "BOTTOMLEFT", "l": 107.21139526367188, "r": 504.4731750488281, "t": 414.0393371582031 }, "charspan": [ 0, 429 ], "page_no": 6 } ], "self_ref": "#/texts/74", "text": "where pos is the position and i is the dimension. That is, each dimension of the positional encoding corresponds to a sinusoid. The wavelengths form a geometric progression from 2 π to 10000 · 2 π . We chose this function because we hypothesized it would allow the model to easily learn to attend by relative positions, since for any fixed offset k , PE$_{pos}$$_{+}$$_{k}$ can be represented as a linear function of PE$_{pos}$ ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We also experimented with using learned positional embeddings [9] instead, and found that the two versions produced nearly identical results (see Table 3 row (E)). We chose the sinusoidal version because it may allow the model to extrapolate to sequence lengths longer than the ones encountered during training.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 311.5639953613281, "coord_origin": "BOTTOMLEFT", "l": 106.97831726074219, "r": 504.12213134765625, "t": 354.3194274902344 }, "charspan": [ 0, 311 ], "page_no": 6 } ], "self_ref": "#/texts/75", "text": "We also experimented with using learned positional embeddings [9] instead, and found that the two versions produced nearly identical results (see Table 3 row (E)). We chose the sinusoidal version because it may allow the model to extrapolate to sequence lengths longer than the ones encountered during training." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "4 Why Self-Attention", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 283.2493591308594, "coord_origin": "BOTTOMLEFT", "l": 107.2066879272461, "r": 225.04141235351562, "t": 295.08807373046875 }, "charspan": [ 0, 20 ], "page_no": 6 } ], "self_ref": "#/texts/76", "text": "4 Why Self-Attention" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In this section we compare various aspects of self-attention layers to the recurrent and convolutional layers commonly used for mapping one variable-length sequence of symbol representations ( x$_{1}$, ..., x$_{n}$ ) to another sequence of equal length ( z$_{1}$, ..., z$_{n}$ ) , with x$_{i}$, z$_{i}$ ∈ R $^{d}$, such as a hidden layer in a typical sequence transduction encoder or decoder. Motivating our use of self-attention we consider three desiderata.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 217.19508361816406, "coord_origin": "BOTTOMLEFT", "l": 106.8187255859375, "r": 505.6536865234375, "t": 270.29833984375 }, "charspan": [ 0, 459 ], "page_no": 6 } ], "self_ref": "#/texts/77", "text": "In this section we compare various aspects of self-attention layers to the recurrent and convolutional layers commonly used for mapping one variable-length sequence of symbol representations ( x$_{1}$, ..., x$_{n}$ ) to another sequence of equal length ( z$_{1}$, ..., z$_{n}$ ) , with x$_{i}$, z$_{i}$ ∈ R $^{d}$, such as a hidden layer in a typical sequence transduction encoder or decoder. Motivating our use of self-attention we consider three desiderata." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "One is the total computational complexity per layer. Another is the amount of computation that can be parallelized, as measured by the minimum number of sequential operations required.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 189.5093994140625, "coord_origin": "BOTTOMLEFT", "l": 107.16207122802734, "r": 504.0003967285156, "t": 210.31512451171875 }, "charspan": [ 0, 184 ], "page_no": 6 } ], "self_ref": "#/texts/78", "text": "One is the total computational complexity per layer. Another is the amount of computation that can be parallelized, as measured by the minimum number of sequential operations required." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The third is the path length between long-range dependencies in the network. Learning long-range dependencies is a key challenge in many sequence transduction tasks. One key factor affecting the ability to learn such dependencies is the length of the paths forward and backward signals have to traverse in the network. The shorter these paths between any combination of positions in the input and output sequences, the easier it is to learn long-range dependencies [12]. Hence we also compare the maximum path length between any two input and output positions in networks composed of the different layer types.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 107.40753173828125, "coord_origin": "BOTTOMLEFT", "l": 107.04078674316406, "r": 504.02520751953125, "t": 182.79638671875 }, "charspan": [ 0, 610 ], "page_no": 6 } ], "self_ref": "#/texts/79", "text": "The third is the path length between long-range dependencies in the network. Learning long-range dependencies is a key challenge in many sequence transduction tasks. One key factor affecting the ability to learn such dependencies is the length of the paths forward and backward signals have to traverse in the network. The shorter these paths between any combination of positions in the input and output sequences, the easier it is to learn long-range dependencies [12]. Hence we also compare the maximum path length between any two input and output positions in networks composed of the different layer types." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As noted in Table 1, a self-attention layer connects all positions with a constant number of sequentially executed operations, whereas a recurrent layer requires O ( n ) sequential operations. In terms of computational complexity, self-attention layers are faster than recurrent layers when the sequence", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.35009765625, "coord_origin": "BOTTOMLEFT", "l": 107.15330505371094, "r": 504.56842041015625, "t": 101.541015625 }, "charspan": [ 0, 303 ], "page_no": 6 } ], "self_ref": "#/texts/80", "text": "As noted in Table 1, a self-attention layer connects all positions with a constant number of sequentially executed operations, whereas a recurrent layer requires O ( n ) sequential operations. In terms of computational complexity, self-attention layers are faster than recurrent layers when the sequence" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "6", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.6934509277344, "r": 308.5073547363281, "t": 49.5316162109375 }, "charspan": [ 0, 1 ], "page_no": 6 } ], "self_ref": "#/texts/81", "text": "6" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "length n is smaller than the representation dimensionality d , which is most often the case with sentence representations used by state-of-the-art models in machine translations, such as word-piece [38] and byte-pair [31] representations. To improve computational performance for tasks involving very long sequences, self-attention could be restricted to considering only a neighborhood of size r in the input sequence centered around the respective output position. This would increase the maximum path length to O ( n/r ) . We plan to investigate this approach further in future work.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 652.7548828125, "coord_origin": "BOTTOMLEFT", "l": 107.09317016601562, "r": 504.20416259765625, "t": 717.6653442382812 }, "charspan": [ 0, 586 ], "page_no": 7 } ], "self_ref": "#/texts/82", "text": "length n is smaller than the representation dimensionality d , which is most often the case with sentence representations used by state-of-the-art models in machine translations, such as word-piece [38] and byte-pair [31] representations. To improve computational performance for tasks involving very long sequences, self-attention could be restricted to considering only a neighborhood of size r in the input sequence centered around the respective output position. This would increase the maximum path length to O ( n/r ) . We plan to investigate this approach further in future work." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "A single convolutional layer with kernel width k < n does not connect all pairs of input and output positions. Doing so requires a stack of O ( n/k ) convolutional layers in the case of contiguous kernels, or O ( log$_{k}$ ( n )) in the case of dilated convolutions [18], increasing the length of the longest paths between any two positions in the network. Convolutional layers are generally more expensive than recurrent layers, by a factor of k . Separable convolutions [6], however, decrease the complexity considerably, to O ( k · n · d + n · d $^{2}$) . Even with k = n , however, the complexity of a separable convolution is equal to the combination of a self-attention layer and a point-wise feed-forward layer, the approach we take in our model.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 560.2037353515625, "coord_origin": "BOTTOMLEFT", "l": 107.02108764648438, "r": 505.24774169921875, "t": 646.5534057617188 }, "charspan": [ 0, 753 ], "page_no": 7 } ], "self_ref": "#/texts/83", "text": "A single convolutional layer with kernel width k < n does not connect all pairs of input and output positions. Doing so requires a stack of O ( n/k ) convolutional layers in the case of contiguous kernels, or O ( log$_{k}$ ( n )) in the case of dilated convolutions [18], increasing the length of the longest paths between any two positions in the network. Convolutional layers are generally more expensive than recurrent layers, by a factor of k . Separable convolutions [6], however, decrease the complexity considerably, to O ( k · n · d + n · d $^{2}$) . Even with k = n , however, the complexity of a separable convolution is equal to the combination of a self-attention layer and a point-wise feed-forward layer, the approach we take in our model." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As side benefit, self-attention could yield more interpretable models. We inspect attention distributions from our models and present and discuss examples in the appendix. Not only do individual attention heads clearly learn to perform different tasks, many appear to exhibit behavior related to the syntactic and semantic structure of the sentences.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 511.47210693359375, "coord_origin": "BOTTOMLEFT", "l": 107.11595153808594, "r": 504.02947998046875, "t": 554.0537109375 }, "charspan": [ 0, 350 ], "page_no": 7 } ], "self_ref": "#/texts/84", "text": "As side benefit, self-attention could yield more interpretable models. We inspect attention distributions from our models and present and discuss examples in the appendix. Not only do individual attention heads clearly learn to perform different tasks, many appear to exhibit behavior related to the syntactic and semantic structure of the sentences." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5 Training", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 479.90374755859375, "coord_origin": "BOTTOMLEFT", "l": 107.66629028320312, "r": 170.4154052734375, "t": 491.8029479980469 }, "charspan": [ 0, 10 ], "page_no": 7 } ], "self_ref": "#/texts/85", "text": "5 Training" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "This section describes the training regime for our models.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 456.25311279296875, "coord_origin": "BOTTOMLEFT", "l": 106.92620086669922, "r": 337.4783935546875, "t": 466.0622253417969 }, "charspan": [ 0, 58 ], "page_no": 7 } ], "self_ref": "#/texts/86", "text": "This section describes the training regime for our models." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5.1 Training Data and Batching", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 428.89642333984375, "coord_origin": "BOTTOMLEFT", "l": 107.47438049316406, "r": 249.97442626953125, "t": 438.78350830078125 }, "charspan": [ 0, 30 ], "page_no": 7 } ], "self_ref": "#/texts/87", "text": "5.1 Training Data and Batching" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We trained on the standard WMT 2014 English-German dataset consisting of about 4.5 million sentence pairs. Sentences were encoded using byte-pair encoding [3], which has a shared sourcetarget vocabulary of about 37000 tokens. For English-French, we used the significantly larger WMT 2014 English-French dataset consisting of 36M sentences and split tokens into a 32000 word-piece vocabulary [38]. Sentence pairs were batched together by approximate sequence length. Each training batch contained a set of sentence pairs containing approximately 25000 source tokens and 25000 target tokens.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 343.1160888671875, "coord_origin": "BOTTOMLEFT", "l": 106.94161987304688, "r": 505.65435791015625, "t": 418.1543884277344 }, "charspan": [ 0, 589 ], "page_no": 7 } ], "self_ref": "#/texts/88", "text": "We trained on the standard WMT 2014 English-German dataset consisting of about 4.5 million sentence pairs. Sentences were encoded using byte-pair encoding [3], which has a shared sourcetarget vocabulary of about 37000 tokens. For English-French, we used the significantly larger WMT 2014 English-French dataset consisting of 36M sentences and split tokens into a 32000 word-piece vocabulary [38]. Sentence pairs were batched together by approximate sequence length. Each training batch contained a set of sentence pairs containing approximately 25000 source tokens and 25000 target tokens." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5.2 Hardware and Schedule", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 316.4078063964844, "coord_origin": "BOTTOMLEFT", "l": 107.60478973388672, "r": 233.04058837890625, "t": 326.3747253417969 }, "charspan": [ 0, 25 ], "page_no": 7 } ], "self_ref": "#/texts/89", "text": "5.2 Hardware and Schedule" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We trained our models on one machine with 8 NVIDIA P100 GPUs. For our base models using the hyperparameters described throughout the paper, each training step took about 0.4 seconds. We trained the base models for a total of 100,000 steps or 12 hours. For our big models,(described on the bottom line of table 3), step time was 1.0 seconds. The big models were trained for 300,000 steps (3.5 days).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 251.41058349609375, "coord_origin": "BOTTOMLEFT", "l": 106.9512710571289, "r": 504.0013732910156, "t": 304.8434753417969 }, "charspan": [ 0, 398 ], "page_no": 7 } ], "self_ref": "#/texts/90", "text": "We trained our models on one machine with 8 NVIDIA P100 GPUs. For our base models using the hyperparameters described throughout the paper, each training step took about 0.4 seconds. We trained the base models for a total of 100,000 steps or 12 hours. For our big models,(described on the bottom line of table 3), step time was 1.0 seconds. The big models were trained for 300,000 steps (3.5 days)." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5.3 Optimizer", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 224.8634033203125, "coord_origin": "BOTTOMLEFT", "l": 107.5698013305664, "r": 174.13174438476562, "t": 235.13262939453125 }, "charspan": [ 0, 13 ], "page_no": 7 } ], "self_ref": "#/texts/91", "text": "5.3 Optimizer" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We used the Adam optimizer [20] with β$_{1}$ = 0 . 9 , β$_{2}$ = 0 . 98 and ϵ = 10 - $^{9}$. We varied the learning rate over the course of training, according to the formula:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 192.96490478515625, "coord_origin": "BOTTOMLEFT", "l": 107.05850982666016, "r": 504.0029296875, "t": 215.1133575439453 }, "charspan": [ 0, 175 ], "page_no": 7 } ], "self_ref": "#/texts/92", "text": "We used the Adam optimizer [20] with β$_{1}$ = 0 . 9 , β$_{2}$ = 0 . 98 and ϵ = 10 - $^{9}$. We varied the learning rate over the course of training, according to the formula:" }, { "children": [], "enumerated": null, "label": "formula", "level": null, "marker": null, "orig": "lrate = d - 0 . 5 model · min( step _ num - 0 . $^{5}$, step _ num · warmup _ steps - 1 . $^{5}$) (3)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 160.57064819335938, "coord_origin": "BOTTOMLEFT", "l": 162.37283325195312, "r": 504.7447204589844, "t": 175.5289306640625 }, "charspan": [ 0, 101 ], "page_no": 7 } ], "self_ref": "#/texts/93", "text": "lrate = d - 0 . 5 model · min( step _ num - 0 . $^{5}$, step _ num · warmup _ steps - 1 . $^{5}$) (3)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "This corresponds to increasing the learning rate linearly for the first warmup _ steps training steps, and decreasing it thereafter proportionally to the inverse square root of the step number. We used warmup _ steps = 4000 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 117.53008270263672, "coord_origin": "BOTTOMLEFT", "l": 107.29002380371094, "r": 505.2438049316406, "t": 148.61663818359375 }, "charspan": [ 0, 225 ], "page_no": 7 } ], "self_ref": "#/texts/94", "text": "This corresponds to increasing the learning rate linearly for the first warmup _ steps training steps, and decreasing it thereafter proportionally to the inverse square root of the step number. We used warmup _ steps = 4000 ." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5.4 Regularization", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 90.33038330078125, "coord_origin": "BOTTOMLEFT", "l": 107.50627136230469, "r": 193.50897216796875, "t": 100.1263427734375 }, "charspan": [ 0, 18 ], "page_no": 7 } ], "self_ref": "#/texts/95", "text": "5.4 Regularization" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We employ three types of regularization during training:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.693115234375, "coord_origin": "BOTTOMLEFT", "l": 106.96537017822266, "r": 331.9893493652344, "t": 79.29132080078125 }, "charspan": [ 0, 56 ], "page_no": 7 } ], "self_ref": "#/texts/96", "text": "We employ three types of regularization during training:" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "7", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.84710693359375, "r": 308.49029541015625, "t": 49.471923828125 }, "charspan": [ 0, 1 ], "page_no": 7 } ], "self_ref": "#/texts/97", "text": "7" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Table 2: The Transformer achieves better BLEU scores than previous state-of-the-art models on the English-to-German and English-to-French newstest2014 tests at a fraction of the training cost.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 699.9127197265625, "coord_origin": "BOTTOMLEFT", "l": 106.9210433959961, "r": 504.0032043457031, "t": 720.7218017578125 }, "charspan": [ 0, 192 ], "page_no": 8 } ], "self_ref": "#/texts/98", "text": "Table 2: The Transformer achieves better BLEU scores than previous state-of-the-art models on the English-to-German and English-to-French newstest2014 tests at a fraction of the training cost." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Residual Dropout We apply dropout [33] to the output of each sub-layer, before it is added to the sub-layer input and normalized. In addition, we apply dropout to the sums of the embeddings and the positional encodings in both the encoder and decoder stacks. For the base model, we use a rate of P$_{drop}$ = 0 . 1 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 475.3942565917969, "coord_origin": "BOTTOMLEFT", "l": 107.15463256835938, "r": 504.2806396484375, "t": 519.0767211914062 }, "charspan": [ 0, 316 ], "page_no": 8 } ], "self_ref": "#/texts/99", "text": "Residual Dropout We apply dropout [33] to the output of each sub-layer, before it is added to the sub-layer input and normalized. In addition, we apply dropout to the sums of the embeddings and the positional encodings in both the encoder and decoder stacks. For the base model, we use a rate of P$_{drop}$ = 0 . 1 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Label Smoothing During training, we employed label smoothing of value ϵ$_{ls}$ = 0 . 1 [36]. This hurts perplexity, as the model learns to be more unsure, but improves accuracy and BLEU score.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 439.4449768066406, "coord_origin": "BOTTOMLEFT", "l": 107.3036117553711, "r": 504.0899963378906, "t": 460.7943420410156 }, "charspan": [ 0, 192 ], "page_no": 8 } ], "self_ref": "#/texts/100", "text": "Label Smoothing During training, we employed label smoothing of value ϵ$_{ls}$ = 0 . 1 [36]. This hurts perplexity, as the model learns to be more unsure, but improves accuracy and BLEU score." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "6 Results", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 409.03936767578125, "coord_origin": "BOTTOMLEFT", "l": 107.46460723876953, "r": 163.2405548095703, "t": 420.1395568847656 }, "charspan": [ 0, 9 ], "page_no": 8 } ], "self_ref": "#/texts/101", "text": "6 Results" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "6.1 Machine Translation", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 385.0038146972656, "coord_origin": "BOTTOMLEFT", "l": 107.52304077148438, "r": 219.07301330566406, "t": 395.06793212890625 }, "charspan": [ 0, 23 ], "page_no": 8 } ], "self_ref": "#/texts/102", "text": "6.1 Machine Translation" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "On the WMT 2014 English-to-German translation task, the big transformer model (Transformer (big) in Table 2) outperforms the best previously reported models (including ensembles) by more than 2 . 0 BLEU, establishing a new state-of-the-art BLEU score of 28 . 4 . The configuration of this model is listed in the bottom line of Table 3. Training took 3 . 5 days on 8 P100 GPUs. Even our base model surpasses all previously published models and ensembles, at a fraction of the training cost of any of the competitive models.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 309.5210876464844, "coord_origin": "BOTTOMLEFT", "l": 107.20944213867188, "r": 504.6653137207031, "t": 373.7106628417969 }, "charspan": [ 0, 522 ], "page_no": 8 } ], "self_ref": "#/texts/103", "text": "On the WMT 2014 English-to-German translation task, the big transformer model (Transformer (big) in Table 2) outperforms the best previously reported models (including ensembles) by more than 2 . 0 BLEU, establishing a new state-of-the-art BLEU score of 28 . 4 . The configuration of this model is listed in the bottom line of Table 3. Training took 3 . 5 days on 8 P100 GPUs. Even our base model surpasses all previously published models and ensembles, at a fraction of the training cost of any of the competitive models." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "On the WMT 2014 English-to-French translation task, our big model achieves a BLEU score of 41 . 0 , outperforming all of the previously published single models, at less than 1 / 4 the training cost of the previous state-of-the-art model. The Transformer (big) model trained for English-to-French used dropout rate P$_{drop}$ = 0 . 1 , instead of 0 . 3 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 260.009765625, "coord_origin": "BOTTOMLEFT", "l": 107.1940689086914, "r": 505.2453308105469, "t": 302.983154296875 }, "charspan": [ 0, 353 ], "page_no": 8 } ], "self_ref": "#/texts/104", "text": "On the WMT 2014 English-to-French translation task, our big model achieves a BLEU score of 41 . 0 , outperforming all of the previously published single models, at less than 1 / 4 the training cost of the previous state-of-the-art model. The Transformer (big) model trained for English-to-French used dropout rate P$_{drop}$ = 0 . 1 , instead of 0 . 3 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "For the base models, we used a single model obtained by averaging the last 5 checkpoints, which were written at 10-minute intervals. For the big models, we averaged the last 20 checkpoints. We used beam search with a beam size of 4 and length penalty α = 0 . 6 [38]. These hyperparameters were chosen after experimentation on the development set. We set the maximum output length during inference to input length + 50 , but terminate early when possible [38].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 200.29852294921875, "coord_origin": "BOTTOMLEFT", "l": 107.14618682861328, "r": 504.1119689941406, "t": 253.73150634765625 }, "charspan": [ 0, 459 ], "page_no": 8 } ], "self_ref": "#/texts/105", "text": "For the base models, we used a single model obtained by averaging the last 5 checkpoints, which were written at 10-minute intervals. For the big models, we averaged the last 20 checkpoints. We used beam search with a beam size of 4 and length penalty α = 0 . 6 [38]. These hyperparameters were chosen after experimentation on the development set. We set the maximum output length during inference to input length + 50 , but terminate early when possible [38]." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Table 2 summarizes our results and compares our translation quality and training costs to other model architectures from the literature. We estimate the number of floating point operations used to train a model by multiplying the training time, the number of GPUs used, and an estimate of the sustained single-precision floating-point capacity of each GPU $^{5}$.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 150.82781982421875, "coord_origin": "BOTTOMLEFT", "l": 106.97612762451172, "r": 504.0679016113281, "t": 193.6265869140625 }, "charspan": [ 0, 363 ], "page_no": 8 } ], "self_ref": "#/texts/106", "text": "Table 2 summarizes our results and compares our translation quality and training costs to other model architectures from the literature. We estimate the number of floating point operations used to train a model by multiplying the training time, the number of GPUs used, and an estimate of the sustained single-precision floating-point capacity of each GPU $^{5}$." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "6.2 Model Variations", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 124.64582824707031, "coord_origin": "BOTTOMLEFT", "l": 107.57784271240234, "r": 204.18380737304688, "t": 134.5999755859375 }, "charspan": [ 0, 20 ], "page_no": 8 } ], "self_ref": "#/texts/107", "text": "6.2 Model Variations" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "To evaluate the importance of different components of the Transformer, we varied our base model in different ways, measuring the change in performance on English-to-German translation on the", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 91.68536376953125, "coord_origin": "BOTTOMLEFT", "l": 107.07888793945312, "r": 504.0005798339844, "t": 113.67156982421875 }, "charspan": [ 0, 190 ], "page_no": 8 } ], "self_ref": "#/texts/108", "text": "To evaluate the importance of different components of the Transformer, we varied our base model in different ways, measuring the change in performance on English-to-German translation on the" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{5}$We used values of 2.8, 3.7, 6.0 and 9.5 TFLOPS for K80, K40, M40 and P100, respectively.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.8465576171875, "coord_origin": "BOTTOMLEFT", "l": 120.315673828125, "r": 453.5559997558594, "t": 79.3341064453125 }, "charspan": [ 0, 94 ], "page_no": 8 } ], "self_ref": "#/texts/109", "text": "$^{5}$We used values of 2.8, 3.7, 6.0 and 9.5 TFLOPS for K80, K40, M40 and P100, respectively." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "8", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.8262023925781, "r": 308.49029541015625, "t": 49.37261962890625 }, "charspan": [ 0, 1 ], "page_no": 8 } ], "self_ref": "#/texts/110", "text": "8" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Table 3: Variations on the Transformer architecture. Unlisted values are identical to those of the base model. All metrics are on the English-to-German translation development set, newstest2013. Listed perplexities are per-wordpiece, according to our byte-pair encoding, and should not be compared to per-word perplexities.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 677.8555908203125, "coord_origin": "BOTTOMLEFT", "l": 106.98860168457031, "r": 504.1953125, "t": 720.727783203125 }, "charspan": [ 0, 323 ], "page_no": 9 } ], "self_ref": "#/texts/111", "text": "Table 3: Variations on the Transformer architecture. Unlisted values are identical to those of the base model. All metrics are on the English-to-German translation development set, newstest2013. Listed perplexities are per-wordpiece, according to our byte-pair encoding, and should not be compared to per-word perplexities." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "development set, newstest2013. We used beam search as described in the previous section, but no checkpoint averaging. We present these results in Table 3.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 357.5391540527344, "coord_origin": "BOTTOMLEFT", "l": 107.44959259033203, "r": 504.07562255859375, "t": 378.2555847167969 }, "charspan": [ 0, 154 ], "page_no": 9 } ], "self_ref": "#/texts/112", "text": "development set, newstest2013. We used beam search as described in the previous section, but no checkpoint averaging. We present these results in Table 3." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In Table 3 rows (A), we vary the number of attention heads and the attention key and value dimensions, keeping the amount of computation constant, as described in Section 3.2.2. While single-head attention is 0.9 BLEU worse than the best setting, quality also drops off with too many heads.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 319.17095947265625, "coord_origin": "BOTTOMLEFT", "l": 107.34579467773438, "r": 505.24127197265625, "t": 350.974609375 }, "charspan": [ 0, 290 ], "page_no": 9 } ], "self_ref": "#/texts/113", "text": "In Table 3 rows (A), we vary the number of attention heads and the attention key and value dimensions, keeping the amount of computation constant, as described in Section 3.2.2. While single-head attention is 0.9 BLEU worse than the best setting, quality also drops off with too many heads." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In Table 3 rows (B), we observe that reducing the attention key size d$_{k}$ hurts model quality. This suggests that determining compatibility is not easy and that a more sophisticated compatibility function than dot product may be beneficial. We further observe in rows (C) and (D) that, as expected, bigger models are better, and dropout is very helpful in avoiding over-fitting. In row (E) we replace our sinusoidal positional encoding with learned positional embeddings [9], and observe nearly identical results to the base model.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 248.49208068847656, "coord_origin": "BOTTOMLEFT", "l": 107.42291259765625, "r": 505.24127197265625, "t": 312.8634338378906 }, "charspan": [ 0, 534 ], "page_no": 9 } ], "self_ref": "#/texts/114", "text": "In Table 3 rows (B), we observe that reducing the attention key size d$_{k}$ hurts model quality. This suggests that determining compatibility is not easy and that a more sophisticated compatibility function than dot product may be beneficial. We further observe in rows (C) and (D) that, as expected, bigger models are better, and dropout is very helpful in avoiding over-fitting. In row (E) we replace our sinusoidal positional encoding with learned positional embeddings [9], and observe nearly identical results to the base model." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "6.3 English Constituency Parsing", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 221.357177734375, "coord_origin": "BOTTOMLEFT", "l": 107.45710754394531, "r": 256.3836364746094, "t": 231.6878662109375 }, "charspan": [ 0, 32 ], "page_no": 9 } ], "self_ref": "#/texts/115", "text": "6.3 English Constituency Parsing" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "To evaluate if the Transformer can generalize to other tasks we performed experiments on English constituency parsing. This task presents specific challenges: the output is subject to strong structural constraints and is significantly longer than the input. Furthermore, RNN sequence-to-sequence models have not been able to attain state-of-the-art results in small-data regimes [37].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 168.01385498046875, "coord_origin": "BOTTOMLEFT", "l": 107.06143188476562, "r": 504.03448486328125, "t": 210.1285400390625 }, "charspan": [ 0, 384 ], "page_no": 9 } ], "self_ref": "#/texts/116", "text": "To evaluate if the Transformer can generalize to other tasks we performed experiments on English constituency parsing. This task presents specific challenges: the output is subject to strong structural constraints and is significantly longer than the input. Furthermore, RNN sequence-to-sequence models have not been able to attain state-of-the-art results in small-data regimes [37]." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We trained a 4-layer transformer with d$_{model}$ = 1024 on the Wall Street Journal (WSJ) portion of the Penn Treebank [25], about 40K training sentences. We also trained it in a semi-supervised setting, using the larger high-confidence and BerkleyParser corpora from with approximately 17M sentences [37]. We used a vocabulary of 16K tokens for the WSJ only setting and a vocabulary of 32K tokens for the semi-supervised setting.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 107.59417724609375, "coord_origin": "BOTTOMLEFT", "l": 106.98470306396484, "r": 505.24371337890625, "t": 161.494140625 }, "charspan": [ 0, 430 ], "page_no": 9 } ], "self_ref": "#/texts/117", "text": "We trained a 4-layer transformer with d$_{model}$ = 1024 on the Wall Street Journal (WSJ) portion of the Penn Treebank [25], about 40K training sentences. We also trained it in a semi-supervised setting, using the larger high-confidence and BerkleyParser corpora from with approximately 17M sentences [37]. We used a vocabulary of 16K tokens for the WSJ only setting and a vocabulary of 32K tokens for the semi-supervised setting." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We performed only a small number of experiments to select the dropout, both attention and residual (section 5.4), learning rates and beam size on the Section 22 development set, all other parameters remained unchanged from the English-to-German base translation model. During inference, we", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.602783203125, "coord_origin": "BOTTOMLEFT", "l": 106.95533752441406, "r": 504.0781555175781, "t": 101.5164794921875 }, "charspan": [ 0, 289 ], "page_no": 9 } ], "self_ref": "#/texts/118", "text": "We performed only a small number of experiments to select the dropout, both attention and residual (section 5.4), learning rates and beam size on the Section 22 development set, all other parameters remained unchanged from the English-to-German base translation model. During inference, we" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "9", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 302.4856262207031, "r": 308.49029541015625, "t": 49.392578125 }, "charspan": [ 0, 1 ], "page_no": 9 } ], "self_ref": "#/texts/119", "text": "9" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Table 4: The Transformer generalizes well to English constituency parsing (Results are on Section 23 of WSJ)", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 700.18408203125, "coord_origin": "BOTTOMLEFT", "l": 106.71479797363281, "r": 503.99493408203125, "t": 720.715087890625 }, "charspan": [ 0, 108 ], "page_no": 10 } ], "self_ref": "#/texts/120", "text": "Table 4: The Transformer generalizes well to English constituency parsing (Results are on Section 23 of WSJ)" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "increased the maximum output length to input length + 300 . We used a beam size of 21 and α = 0 . 3 for both WSJ only and the semi-supervised setting.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 509.0591125488281, "coord_origin": "BOTTOMLEFT", "l": 107.46659088134766, "r": 504.2492980957031, "t": 530.1951904296875 }, "charspan": [ 0, 150 ], "page_no": 10 } ], "self_ref": "#/texts/121", "text": "increased the maximum output length to input length + 300 . We used a beam size of 21 and α = 0 . 3 for both WSJ only and the semi-supervised setting." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Our results in Table 4 show that despite the lack of task-specific tuning our model performs surprisingly well, yielding better results than all previously reported models with the exception of the Recurrent Neural Network Grammar [8].", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 471.0895690917969, "coord_origin": "BOTTOMLEFT", "l": 107.44959259033203, "r": 505.6535949707031, "t": 502.9370422363281 }, "charspan": [ 0, 235 ], "page_no": 10 } ], "self_ref": "#/texts/122", "text": "Our results in Table 4 show that despite the lack of task-specific tuning our model performs surprisingly well, yielding better results than all previously reported models with the exception of the Recurrent Neural Network Grammar [8]." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In contrast to RNN sequence-to-sequence models [37], the Transformer outperforms the BerkeleyParser [29] even when training only on the WSJ training set of 40K sentences.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 443.67938232421875, "coord_origin": "BOTTOMLEFT", "l": 107.42334747314453, "r": 505.67303466796875, "t": 464.0172119140625 }, "charspan": [ 0, 170 ], "page_no": 10 } ], "self_ref": "#/texts/123", "text": "In contrast to RNN sequence-to-sequence models [37], the Transformer outperforms the BerkeleyParser [29] even when training only on the WSJ training set of 40K sentences." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "7 Conclusion", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 415.5043640136719, "coord_origin": "BOTTOMLEFT", "l": 107.21891021728516, "r": 183.06671142578125, "t": 427.0397644042969 }, "charspan": [ 0, 12 ], "page_no": 10 } ], "self_ref": "#/texts/124", "text": "7 Conclusion" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "In this work, we presented the Transformer, the first sequence transduction model based entirely on attention, replacing the recurrent layers most commonly used in encoder-decoder architectures with multi-headed self-attention.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 371.3680725097656, "coord_origin": "BOTTOMLEFT", "l": 107.35597229003906, "r": 503.9967956542969, "t": 402.7541198730469 }, "charspan": [ 0, 227 ], "page_no": 10 } ], "self_ref": "#/texts/125", "text": "In this work, we presented the Transformer, the first sequence transduction model based entirely on attention, replacing the recurrent layers most commonly used in encoder-decoder architectures with multi-headed self-attention." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "For translation tasks, the Transformer can be trained significantly faster than architectures based on recurrent or convolutional layers. On both WMT 2014 English-to-German and WMT 2014 English-to-French translation tasks, we achieve a new state of the art. In the former task our best model outperforms even all previously reported ensembles.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 322.2520751953125, "coord_origin": "BOTTOMLEFT", "l": 107.24177551269531, "r": 503.997314453125, "t": 364.55279541015625 }, "charspan": [ 0, 343 ], "page_no": 10 } ], "self_ref": "#/texts/126", "text": "For translation tasks, the Transformer can be trained significantly faster than architectures based on recurrent or convolutional layers. On both WMT 2014 English-to-German and WMT 2014 English-to-French translation tasks, we achieve a new state of the art. In the former task our best model outperforms even all previously reported ensembles." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "We are excited about the future of attention-based models and plan to apply them to other tasks. We plan to extend the Transformer to problems involving input and output modalities other than text and to investigate local, restricted attention mechanisms to efficiently handle large inputs and outputs such as images, audio and video. Making generation less sequential is another research goals of ours.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 273.1360778808594, "coord_origin": "BOTTOMLEFT", "l": 107.06587982177734, "r": 505.60797119140625, "t": 315.4392395019531 }, "charspan": [ 0, 403 ], "page_no": 10 } ], "self_ref": "#/texts/127", "text": "We are excited about the future of attention-based models and plan to apply them to other tasks. We plan to extend the Transformer to problems involving input and output modalities other than text and to investigate local, restricted attention mechanisms to efficiently handle large inputs and outputs such as images, audio and video. Making generation less sequential is another research goals of ours." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The code we used to train and evaluate our models is available at https://github.com/ tensorflow/tensor2tensor .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 245.779296875, "coord_origin": "BOTTOMLEFT", "l": 106.94966125488281, "r": 505.0509948730469, "t": 266.1514892578125 }, "charspan": [ 0, 112 ], "page_no": 10 } ], "self_ref": "#/texts/128", "text": "The code we used to train and evaluate our models is available at https://github.com/ tensorflow/tensor2tensor ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Acknowledgements We are grateful to Nal Kalchbrenner and Stephan Gouws for their fruitful comments, corrections and inspiration.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 212.2650909423828, "coord_origin": "BOTTOMLEFT", "l": 107.154541015625, "r": 504.0024108886719, "t": 232.8948974609375 }, "charspan": [ 0, 128 ], "page_no": 10 } ], "self_ref": "#/texts/129", "text": "Acknowledgements We are grateful to Nal Kalchbrenner and Stephan Gouws for their fruitful comments, corrections and inspiration." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "References", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 183.9723663330078, "coord_origin": "BOTTOMLEFT", "l": 107.72453308105469, "r": 163.79808044433594, "t": 195.63629150390625 }, "charspan": [ 0, 10 ], "page_no": 10 } ], "self_ref": "#/texts/130", "text": "References" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "[1] Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. Layer normalization. arXiv preprint arXiv:1607.06450 , 2016.", "parent": { "$ref": "#/groups/1" }, "prov": [ { "bbox": { "b": 156.22508239746094, "coord_origin": "BOTTOMLEFT", "l": 112.64569091796875, "r": 504.4076232910156, "t": 176.72576904296875 }, "charspan": [ 0, 119 ], "page_no": 10 } ], "self_ref": "#/texts/131", "text": "[1] Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. Layer normalization. arXiv preprint arXiv:1607.06450 , 2016." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "[2] Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. Neural machine translation by jointly learning to align and translate. CoRR , abs/1409.0473, 2014.", "parent": { "$ref": "#/groups/1" }, "prov": [ { "bbox": { "b": 127.18450927734375, "coord_origin": "BOTTOMLEFT", "l": 112.9466323852539, "r": 504.34503173828125, "t": 148.23834228515625 }, "charspan": [ 0, 154 ], "page_no": 10 } ], "self_ref": "#/texts/132", "text": "[2] Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. Neural machine translation by jointly learning to align and translate. CoRR , abs/1409.0473, 2014." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "[3] Denny Britz, Anna Goldie, Minh-Thang Luong, and Quoc V. Le. Massive exploration of neural machine translation architectures. CoRR , abs/1703.03906, 2017.", "parent": { "$ref": "#/groups/1" }, "prov": [ { "bbox": { "b": 98.64007568359375, "coord_origin": "BOTTOMLEFT", "l": 112.6026840209961, "r": 504.0042419433594, "t": 119.52862548828125 }, "charspan": [ 0, 157 ], "page_no": 10 } ], "self_ref": "#/texts/133", "text": "[3] Denny Britz, Anna Goldie, Minh-Thang Luong, and Quoc V. Le. Massive exploration of neural machine translation architectures. CoRR , abs/1703.03906, 2017." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "[4] Jianpeng Cheng, Li Dong, and Mirella Lapata. Long short-term memory-networks for machine reading. arXiv preprint arXiv:1601.06733 , 2016.", "parent": { "$ref": "#/groups/1" }, "prov": [ { "bbox": { "b": 69.84539794921875, "coord_origin": "BOTTOMLEFT", "l": 112.68844604492188, "r": 504.0023498535156, "t": 90.0672607421875 }, "charspan": [ 0, 141 ], "page_no": 10 } ], "self_ref": "#/texts/134", "text": "[4] Jianpeng Cheng, Li Dong, and Mirella Lapata. Long short-term memory-networks for machine reading. arXiv preprint arXiv:1601.06733 , 2016." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "10", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 301.01898193359375, "r": 311.2342834472656, "t": 49.35711669921875 }, "charspan": [ 0, 2 ], "page_no": 10 } ], "self_ref": "#/texts/135", "text": "10" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "11", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.9831848144531, "r": 310.9815673828125, "t": 49.42071533203125 }, "charspan": [ 0, 2 ], "page_no": 11 } ], "self_ref": "#/texts/136", "text": "11" }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "12", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 300.9915771484375, "r": 311.0081481933594, "t": 49.70880126953125 }, "charspan": [ 0, 2 ], "page_no": 12 } ], "self_ref": "#/texts/137", "text": "12" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Attention Visualizations Input-Input Layer5", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 707.538330078125, "coord_origin": "BOTTOMLEFT", "l": 107.3423843383789, "r": 250.4954071044922, "t": 731.3789672851562 }, "charspan": [ 0, 43 ], "page_no": 13 } ], "self_ref": "#/texts/138", "text": "Attention Visualizations Input-Input Layer5" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 3: An example of the attention mechanism following long-distance dependencies in the encoder self-attention in layer 5 of 6. Many of the attention heads attend to a distant dependency of the verb 'making', completing the phrase 'making...more difficult'. Attentions here shown only for the word 'making'. Different colors represent different heads. Best viewed in color.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 437.0728454589844, "coord_origin": "BOTTOMLEFT", "l": 107.1154556274414, "r": 504.4945983886719, "t": 479.7002258300781 }, "charspan": [ 0, 377 ], "page_no": 13 } ], "self_ref": "#/texts/139", "text": "Figure 3: An example of the attention mechanism following long-distance dependencies in the encoder self-attention in layer 5 of 6. Many of the attention heads attend to a distant dependency of the verb 'making', completing the phrase 'making...more difficult'. Attentions here shown only for the word 'making'. Different colors represent different heads. Best viewed in color." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "13", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.960079193115234, "coord_origin": "BOTTOMLEFT", "l": 301.01898193359375, "r": 311.0067138671875, "t": 49.4918212890625 }, "charspan": [ 0, 2 ], "page_no": 13 } ], "self_ref": "#/texts/140", "text": "13" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Input-Input Layer5", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 669.127685546875, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 278.68414306640625, "t": 696.4042358398438 }, "charspan": [ 0, 18 ], "page_no": 14 } ], "self_ref": "#/texts/141", "text": "Input-Input Layer5" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 4: Two attention heads, also in layer 5 of 6, apparently involved in anaphora resolution. Top: Full attentions for head 5. Bottom: Isolated attentions from just the word 'its' for attention heads 5 and 6. Note that the attentions are very sharp for this word.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 146.18414306640625, "coord_origin": "BOTTOMLEFT", "l": 107.25314331054688, "r": 505.38818359375, "t": 177.7108154296875 }, "charspan": [ 0, 266 ], "page_no": 14 } ], "self_ref": "#/texts/142", "text": "Figure 4: Two attention heads, also in layer 5 of 6, apparently involved in anaphora resolution. Top: Full attentions for head 5. Bottom: Isolated attentions from just the word 'its' for attention heads 5 and 6. Note that the attentions are very sharp for this word." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "14", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.96006393432617, "coord_origin": "BOTTOMLEFT", "l": 301.01898193359375, "r": 310.9815673828125, "t": 49.4730224609375 }, "charspan": [ 0, 2 ], "page_no": 14 } ], "self_ref": "#/texts/143", "text": "14" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Input-Input Layer5", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 654.8203735351562, "coord_origin": "BOTTOMLEFT", "l": 106.9734878540039, "r": 278.1088562011719, "t": 682.1690673828125 }, "charspan": [ 0, 18 ], "page_no": 15 } ], "self_ref": "#/texts/144", "text": "Input-Input Layer5" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 5: Many of the attention heads exhibit behaviour that seems related to the structure of the sentence. We give two such examples above, from two different heads from the encoder self-attention at layer 5 of 6. The heads clearly learned to perform different tasks.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 157.27392578125, "coord_origin": "BOTTOMLEFT", "l": 107.3069839477539, "r": 504.1837158203125, "t": 189.67730712890625 }, "charspan": [ 0, 269 ], "page_no": 15 } ], "self_ref": "#/texts/145", "text": "Figure 5: Many of the attention heads exhibit behaviour that seems related to the structure of the sentence. We give two such examples above, from two different heads from the encoder self-attention at layer 5 of 6. The heads clearly learned to perform different tasks." }, { "children": [], "enumerated": null, "label": "page_footer", "level": null, "marker": null, "orig": "15", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 39.9600715637207, "coord_origin": "BOTTOMLEFT", "l": 301.01898193359375, "r": 310.9930419921875, "t": 49.69146728515625 }, "charspan": [ 0, 2 ], "page_no": 15 } ], "self_ref": "#/texts/146", "text": "15" } ]

[ { "annotations": [], "captions": [ { "$ref": "#/texts/31" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 397.27667236328125, "coord_origin": "BOTTOMLEFT", "l": 196.17901611328125, "r": 415.58905029296875, "t": 720 }, "charspan": [ 0, 47 ], "page_no": 3 } ], "references": [], "self_ref": "#/pictures/0" }, { "annotations": [], "captions": [], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 572.2972412109375, "coord_origin": "BOTTOMLEFT", "l": 175.1226806640625, "r": 239.90184020996094, "t": 697.697998046875 }, "charspan": [ 0, 0 ], "page_no": 4 } ], "references": [], "self_ref": "#/pictures/1" }, { "annotations": [], "captions": [ { "$ref": "#/texts/40" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 553.0906982421875, "coord_origin": "BOTTOMLEFT", "l": 346.8466491699219, "r": 467.6957092285156, "t": 720.5292358398438 }, "charspan": [ 0, 133 ], "page_no": 4 } ], "references": [], "self_ref": "#/pictures/2" }, { "annotations": [], "captions": [ { "$ref": "#/texts/139" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 489.8687438964844, "coord_origin": "BOTTOMLEFT", "l": 119.5811538696289, "r": 503.3612976074219, "t": 691.8294067382812 }, "charspan": [ 0, 377 ], "page_no": 13 } ], "references": [], "self_ref": "#/pictures/3" }, { "annotations": [], "captions": [ { "$ref": "#/texts/142" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 186.96783447265625, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 502.0914001464844, "t": 621.9672241210938 }, "charspan": [ 0, 266 ], "page_no": 14 } ], "references": [], "self_ref": "#/pictures/4" }, { "annotations": [], "captions": [ { "$ref": "#/texts/145" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 201.84991455078125, "coord_origin": "BOTTOMLEFT", "l": 106.76576232910156, "r": 499.4664306640625, "t": 607.6978149414062 }, "charspan": [ 0, 269 ], "page_no": 15 } ], "references": [], "self_ref": "#/pictures/5" } ]

[ { "captions": [ { "$ref": "#/texts/69" } ], "children": [], "data": { "grid": [ [ { "bbox": { "b": 665.6541137695312, "coord_origin": "BOTTOMLEFT", "l": 124.5469970703125, "r": 169.94656372070312, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Layer Type" }, { "bbox": { "b": 665.6541137695312, "coord_origin": "BOTTOMLEFT", "l": 239.70468139648438, "r": 327.544921875, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Complexity per Layer" }, { "bbox": { "b": 654.7451171875, "coord_origin": "BOTTOMLEFT", "l": 339.4989929199219, "r": 383.2148742675781, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Sequential Operations" }, { "bbox": { "b": 665.6541137695312, "coord_origin": "BOTTOMLEFT", "l": 395.1710510253906, "r": 487.45465087890625, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "Maximum Path Length" } ], [ { "bbox": { "b": 642.1080932617188, "coord_origin": "BOTTOMLEFT", "l": 124.5469970703125, "r": 181.55299377441406, "t": 651.0146484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Self-Attention" }, { "bbox": { "b": 642.3272705078125, "coord_origin": "BOTTOMLEFT", "l": 264.39599609375, "r": 302.8514709472656, "t": 652.71484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "O ( n 2 · d )" }, { "bbox": { "b": 642.3272705078125, "coord_origin": "BOTTOMLEFT", "l": 351.05401611328125, "r": 371.6602478027344, "t": 651.174072265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "O (1)" }, { "bbox": { "b": 642.3272705078125, "coord_origin": "BOTTOMLEFT", "l": 431.0080261230469, "r": 451.6142578125, "t": 651.174072265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "O (1)" } ], [ { "bbox": { "b": 630.72509765625, "coord_origin": "BOTTOMLEFT", "l": 124.54702758789062, "r": 163.82955932617188, "t": 639.6316528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Recurrent" }, { "bbox": { "b": 630.9442749023438, "coord_origin": "BOTTOMLEFT", "l": 264.3960266113281, "r": 302.8514709472656, "t": 640.3489379882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "O ( n · d $^{2}$)" }, { "bbox": { "b": 630.9442749023438, "coord_origin": "BOTTOMLEFT", "l": 350.55401611328125, "r": 372.15948486328125, "t": 639.7910766601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "O ( n )" }, { "bbox": { "b": 630.9442749023438, "coord_origin": "BOTTOMLEFT", "l": 430.509033203125, "r": 452.1134948730469, "t": 639.7910766601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "O ( n )" } ], [ { "bbox": { "b": 619.3430786132812, "coord_origin": "BOTTOMLEFT", "l": 124.54705810546875, "r": 180.9652557373047, "t": 628.2496337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Convolutional" }, { "bbox": { "b": 619.562255859375, "coord_origin": "BOTTOMLEFT", "l": 258.049072265625, "r": 309.198486328125, "t": 628.9669189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "O ( k · n · d $^{2}$)" }, { "bbox": { "b": 619.562255859375, "coord_origin": "BOTTOMLEFT", "l": 351.0540466308594, "r": 371.6602783203125, "t": 628.4090576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "O (1)" }, { "bbox": { "b": 619.562255859375, "coord_origin": "BOTTOMLEFT", "l": 417.8090515136719, "r": 464.81396484375, "t": 628.4090576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "O ( log$_{k}$ ( n ))" } ], [ { "bbox": { "b": 608.43408203125, "coord_origin": "BOTTOMLEFT", "l": 124.54705810546875, "r": 227.7496337890625, "t": 617.3406372070312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Self-Attention (restricted)" }, { "bbox": { "b": 608.6532592773438, "coord_origin": "BOTTOMLEFT", "l": 260.6480712890625, "r": 306.5994873046875, "t": 618.0579223632812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "O ( r · n · d )" }, { "bbox": { "b": 608.6532592773438, "coord_origin": "BOTTOMLEFT", "l": 351.0540466308594, "r": 371.6602783203125, "t": 617.5000610351562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "O (1)" }, { "bbox": { "b": 608.6532592773438, "coord_origin": "BOTTOMLEFT", "l": 425.633056640625, "r": 456.9905090332031, "t": 617.5000610351562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "O ( n/r )" } ] ], "num_cols": 4, "num_rows": 5, "table_cells": [ { "bbox": { "b": 665.6541137695312, "coord_origin": "BOTTOMLEFT", "l": 124.5469970703125, "r": 169.94656372070312, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Layer Type" }, { "bbox": { "b": 665.6541137695312, "coord_origin": "BOTTOMLEFT", "l": 239.70468139648438, "r": 327.544921875, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Complexity per Layer" }, { "bbox": { "b": 654.7451171875, "coord_origin": "BOTTOMLEFT", "l": 339.4989929199219, "r": 383.2148742675781, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Sequential Operations" }, { "bbox": { "b": 665.6541137695312, "coord_origin": "BOTTOMLEFT", "l": 395.1710510253906, "r": 487.45465087890625, "t": 674.5606689453125 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "Maximum Path Length" }, { "bbox": { "b": 642.1080932617188, "coord_origin": "BOTTOMLEFT", "l": 124.5469970703125, "r": 181.55299377441406, "t": 651.0146484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Self-Attention" }, { "bbox": { "b": 642.3272705078125, "coord_origin": "BOTTOMLEFT", "l": 264.39599609375, "r": 302.8514709472656, "t": 652.71484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "O ( n 2 · d )" }, { "bbox": { "b": 642.3272705078125, "coord_origin": "BOTTOMLEFT", "l": 351.05401611328125, "r": 371.6602478027344, "t": 651.174072265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "O (1)" }, { "bbox": { "b": 642.3272705078125, "coord_origin": "BOTTOMLEFT", "l": 431.0080261230469, "r": 451.6142578125, "t": 651.174072265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "O (1)" }, { "bbox": { "b": 630.72509765625, "coord_origin": "BOTTOMLEFT", "l": 124.54702758789062, "r": 163.82955932617188, "t": 639.6316528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Recurrent" }, { "bbox": { "b": 630.9442749023438, "coord_origin": "BOTTOMLEFT", "l": 264.3960266113281, "r": 302.8514709472656, "t": 640.3489379882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "O ( n · d $^{2}$)" }, { "bbox": { "b": 630.9442749023438, "coord_origin": "BOTTOMLEFT", "l": 350.55401611328125, "r": 372.15948486328125, "t": 639.7910766601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "O ( n )" }, { "bbox": { "b": 630.9442749023438, "coord_origin": "BOTTOMLEFT", "l": 430.509033203125, "r": 452.1134948730469, "t": 639.7910766601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "O ( n )" }, { "bbox": { "b": 619.3430786132812, "coord_origin": "BOTTOMLEFT", "l": 124.54705810546875, "r": 180.9652557373047, "t": 628.2496337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Convolutional" }, { "bbox": { "b": 619.562255859375, "coord_origin": "BOTTOMLEFT", "l": 258.049072265625, "r": 309.198486328125, "t": 628.9669189453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "O ( k · n · d $^{2}$)" }, { "bbox": { "b": 619.562255859375, "coord_origin": "BOTTOMLEFT", "l": 351.0540466308594, "r": 371.6602783203125, "t": 628.4090576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "O (1)" }, { "bbox": { "b": 619.562255859375, "coord_origin": "BOTTOMLEFT", "l": 417.8090515136719, "r": 464.81396484375, "t": 628.4090576171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "O ( log$_{k}$ ( n ))" }, { "bbox": { "b": 608.43408203125, "coord_origin": "BOTTOMLEFT", "l": 124.54705810546875, "r": 227.7496337890625, "t": 617.3406372070312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Self-Attention (restricted)" }, { "bbox": { "b": 608.6532592773438, "coord_origin": "BOTTOMLEFT", "l": 260.6480712890625, "r": 306.5994873046875, "t": 618.0579223632812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "O ( r · n · d )" }, { "bbox": { "b": 608.6532592773438, "coord_origin": "BOTTOMLEFT", "l": 351.0540466308594, "r": 371.6602783203125, "t": 617.5000610351562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "O (1)" }, { "bbox": { "b": 608.6532592773438, "coord_origin": "BOTTOMLEFT", "l": 425.633056640625, "r": 456.9905090332031, "t": 617.5000610351562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "O ( n/r )" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 604.8668212890625, "coord_origin": "BOTTOMLEFT", "l": 117.5134506225586, "r": 493.74737548828125, "t": 679.452880859375 }, "charspan": [ 0, 0 ], "page_no": 6 } ], "references": [], "self_ref": "#/tables/0" }, { "captions": [ { "$ref": "#/texts/98" } ], "children": [], "data": { "grid": [ [ { "bbox": { "b": 676.090087890625, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 162.683349609375, "t": 684.9966430664062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Model" }, { "bbox": { "b": 684.3790893554688, "coord_origin": "BOTTOMLEFT", "l": 311.02099609375, "r": 337.0333557128906, "t": 693.28564453125 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "BLEU" }, { "bbox": { "b": 684.3790893554688, "coord_origin": "BOTTOMLEFT", "l": 311.02099609375, "r": 337.0333557128906, "t": 693.28564453125 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "BLEU" }, { "bbox": { "b": 684.3790893554688, "coord_origin": "BOTTOMLEFT", "l": 383.2498474121094, "r": 475.3242492675781, "t": 693.28564453125 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "Training Cost (FLOPs)" }, { "bbox": { "b": 684.3790893554688, "coord_origin": "BOTTOMLEFT", "l": 383.2498474121094, "r": 475.3242492675781, "t": 693.28564453125 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "Training Cost (FLOPs)" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 288.7200012207031, "r": 318.59783935546875, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "EN-DE" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 330.5529479980469, "r": 359.33489990234375, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "EN-FR" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 387.4692687988281, "r": 417.34710693359375, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "EN-DE" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 440.0419006347656, "r": 468.8238525390625, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "EN-FR" } ], [ { "bbox": { "b": 657.1571044921875, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 188.9646759033203, "t": 666.0636596679688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "ByteNet [18]" }, { "bbox": { "b": 657.1571044921875, "coord_origin": "BOTTOMLEFT", "l": 292.4461975097656, "r": 314.862060546875, "t": 666.0636596679688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "23.75" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "" } ], [ { "bbox": { "b": 645.7741088867188, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 234.98193359375, "t": 654.6806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Deep-Att + PosUnk [39]" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 645.7741088867188, "coord_origin": "BOTTOMLEFT", "l": 336.22186279296875, "r": 353.6564025878906, "t": 654.6806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "39.2" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 645.9932861328125, "coord_origin": "BOTTOMLEFT", "l": 435.26397705078125, "r": 473.0951232910156, "t": 656.3818969726562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "1 . 0 · 10 20" } ], [ { "bbox": { "b": 634.3921508789062, "coord_origin": "BOTTOMLEFT", "l": 136.67098999023438, "r": 208.421630859375, "t": 643.2987060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "GNMT + RL [38]" }, { "bbox": { "b": 634.3921508789062, "coord_origin": "BOTTOMLEFT", "l": 294.93682861328125, "r": 312.3713684082031, "t": 643.2987060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "24.6" }, { "bbox": { "b": 634.3921508789062, "coord_origin": "BOTTOMLEFT", "l": 333.7311706542969, "r": 356.14703369140625, "t": 643.2987060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "39.92" }, { "bbox": { "b": 634.611328125, "coord_origin": "BOTTOMLEFT", "l": 383.2449951171875, "r": 421.0761413574219, "t": 644.9989013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "2 . 3 · 10 19" }, { "bbox": { "b": 634.611328125, "coord_origin": "BOTTOMLEFT", "l": 435.2640075683594, "r": 473.09515380859375, "t": 644.9989013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "1 . 4 · 10 20" } ], [ { "bbox": { "b": 623.0091552734375, "coord_origin": "BOTTOMLEFT", "l": 136.6710205078125, "r": 188.02822875976562, "t": 631.9157104492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "ConvS2S [9]" }, { "bbox": { "b": 623.0091552734375, "coord_origin": "BOTTOMLEFT", "l": 292.44622802734375, "r": 314.8620910644531, "t": 631.9157104492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "25.16" }, { "bbox": { "b": 623.0091552734375, "coord_origin": "BOTTOMLEFT", "l": 333.73126220703125, "r": 356.1471252441406, "t": 631.9157104492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "40.46" }, { "bbox": { "b": 623.2283325195312, "coord_origin": "BOTTOMLEFT", "l": 383.2450256347656, "r": 421.076171875, "t": 633.616943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 5, "text": "9 . 6 · 10 18" }, { "bbox": { "b": 623.2283325195312, "coord_origin": "BOTTOMLEFT", "l": 435.2640380859375, "r": 473.0951843261719, "t": 633.616943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 5, "text": "1 . 5 · 10 20" } ], [ { "bbox": { "b": 611.627197265625, "coord_origin": "BOTTOMLEFT", "l": 136.67105102539062, "r": 175.68458557128906, "t": 620.5337524414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "MoE [32]" }, { "bbox": { "b": 611.627197265625, "coord_origin": "BOTTOMLEFT", "l": 292.4462585449219, "r": 314.86212158203125, "t": 620.5337524414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "26.03" }, { "bbox": { "b": 611.627197265625, "coord_origin": "BOTTOMLEFT", "l": 333.7312927246094, "r": 356.14715576171875, "t": 620.5337524414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "40.56" }, { "bbox": { "b": 611.8463745117188, "coord_origin": "BOTTOMLEFT", "l": 383.24505615234375, "r": 421.0762023925781, "t": 622.2339477539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 6, "text": "2 . 0 · 10 19" }, { "bbox": { "b": 611.8463745117188, "coord_origin": "BOTTOMLEFT", "l": 435.2650451660156, "r": 473.0951843261719, "t": 622.2339477539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 6, "text": "1 . 2 · 10 20" } ], [ { "bbox": { "b": 598.9890747070312, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 276.76507568359375, "t": 607.8956298828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Deep-Att + PosUnk Ensemble [39]" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 598.9890747070312, "coord_origin": "BOTTOMLEFT", "l": 336.22186279296875, "r": 353.6564025878906, "t": 607.8956298828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "40.4" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 599.208251953125, "coord_origin": "BOTTOMLEFT", "l": 435.26397705078125, "r": 473.0951232910156, "t": 609.5968627929688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 7, "text": "8 . 0 · 10 20" } ], [ { "bbox": { "b": 587.6071166992188, "coord_origin": "BOTTOMLEFT", "l": 136.67098999023438, "r": 250.20477294921875, "t": 596.513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "GNMT + RL Ensemble [38]" }, { "bbox": { "b": 587.6071166992188, "coord_origin": "BOTTOMLEFT", "l": 292.4461975097656, "r": 314.862060546875, "t": 596.513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "26.30" }, { "bbox": { "b": 587.6071166992188, "coord_origin": "BOTTOMLEFT", "l": 333.7312316894531, "r": 356.1470947265625, "t": 596.513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "41.16" }, { "bbox": { "b": 587.8262939453125, "coord_origin": "BOTTOMLEFT", "l": 383.2449951171875, "r": 421.0761413574219, "t": 598.2138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 8, "text": "1 . 8 · 10 20" }, { "bbox": { "b": 587.8262939453125, "coord_origin": "BOTTOMLEFT", "l": 435.2640075683594, "r": 473.09515380859375, "t": 598.2138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 8, "text": "1 . 1 · 10 21" } ], [ { "bbox": { "b": 576.22509765625, "coord_origin": "BOTTOMLEFT", "l": 136.6710205078125, "r": 229.81137084960938, "t": 585.1316528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "ConvS2S Ensemble [9]" }, { "bbox": { "b": 576.22509765625, "coord_origin": "BOTTOMLEFT", "l": 292.44622802734375, "r": 314.8620910644531, "t": 585.1316528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "26.36" }, { "bbox": { "b": 576.2948608398438, "coord_origin": "BOTTOMLEFT", "l": 333.73602294921875, "r": 356.1518859863281, "t": 585.251220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "41.29" }, { "bbox": { "b": 576.4442749023438, "coord_origin": "BOTTOMLEFT", "l": 383.2450256347656, "r": 421.076171875, "t": 586.8318481445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 9, "text": "7 . 7 · 10 19" }, { "bbox": { "b": 576.4442749023438, "coord_origin": "BOTTOMLEFT", "l": 435.2640380859375, "r": 473.0951843261719, "t": 586.8318481445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 9, "text": "1 . 2 · 10 21" } ], [ { "bbox": { "b": 563.089111328125, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 240.34181213378906, "t": 571.9956665039062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Transformer (base model)" }, { "bbox": { "b": 563.089111328125, "coord_origin": "BOTTOMLEFT", "l": 294.9368591308594, "r": 312.37139892578125, "t": 571.9956665039062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "27.3" }, { "bbox": { "b": 563.089111328125, "coord_origin": "BOTTOMLEFT", "l": 336.22186279296875, "r": 353.6564025878906, "t": 571.9956665039062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "38.1" }, { "bbox": { "b": 563.3082885742188, "coord_origin": "BOTTOMLEFT", "l": 407.3389892578125, "r": 427.7060546875, "t": 572.7129516601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 10, "text": "3 . 3 ·" }, { "bbox": { "b": 563.3082885742188, "coord_origin": "BOTTOMLEFT", "l": 430.25299072265625, "r": 450.73687744140625, "t": 573.6958618164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 10, "text": "10 18" } ], [ { "bbox": { "b": 551.7070922851562, "coord_origin": "BOTTOMLEFT", "l": 136.67098999023438, "r": 207.97329711914062, "t": 560.6136474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Transformer (big)" }, { "bbox": { "b": 551.77685546875, "coord_origin": "BOTTOMLEFT", "l": 294.94097900390625, "r": 312.3755187988281, "t": 560.7332153320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "28.4" }, { "bbox": { "b": 551.77685546875, "coord_origin": "BOTTOMLEFT", "l": 336.2259826660156, "r": 353.6605224609375, "t": 560.7332153320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "41.8" }, { "bbox": { "b": 551.92626953125, "coord_origin": "BOTTOMLEFT", "l": 410.12298583984375, "r": 440.01055908203125, "t": 561.3309326171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 11, "text": "2 . 3 · 10" }, { "bbox": { "b": 556.12109375, "coord_origin": "BOTTOMLEFT", "l": 440.010986328125, "r": 447.9541320800781, "t": 562.3138427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 11, "text": "19" } ] ], "num_cols": 5, "num_rows": 12, "table_cells": [ { "bbox": { "b": 676.090087890625, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 162.683349609375, "t": 684.9966430664062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Model" }, { "bbox": { "b": 684.3790893554688, "coord_origin": "BOTTOMLEFT", "l": 311.02099609375, "r": 337.0333557128906, "t": 693.28564453125 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "BLEU" }, { "bbox": { "b": 684.3790893554688, "coord_origin": "BOTTOMLEFT", "l": 383.2498474121094, "r": 475.3242492675781, "t": 693.28564453125 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "Training Cost (FLOPs)" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 288.7200012207031, "r": 318.59783935546875, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "EN-DE" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 330.5529479980469, "r": 359.33489990234375, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "EN-FR" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 387.4692687988281, "r": 417.34710693359375, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "EN-DE" }, { "bbox": { "b": 668.464111328125, "coord_origin": "BOTTOMLEFT", "l": 440.0419006347656, "r": 468.8238525390625, "t": 677.3706665039062 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "EN-FR" }, { "bbox": { "b": 657.1571044921875, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 188.9646759033203, "t": 666.0636596679688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "ByteNet [18]" }, { "bbox": { "b": 657.1571044921875, "coord_origin": "BOTTOMLEFT", "l": 292.4461975097656, "r": 314.862060546875, "t": 666.0636596679688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "23.75" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "" }, { "bbox": { "b": 645.7741088867188, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 234.98193359375, "t": 654.6806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Deep-Att + PosUnk [39]" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 645.7741088867188, "coord_origin": "BOTTOMLEFT", "l": 336.22186279296875, "r": 353.6564025878906, "t": 654.6806640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "39.2" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 645.9932861328125, "coord_origin": "BOTTOMLEFT", "l": 435.26397705078125, "r": 473.0951232910156, "t": 656.3818969726562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "1 . 0 · 10 20" }, { "bbox": { "b": 634.3921508789062, "coord_origin": "BOTTOMLEFT", "l": 136.67098999023438, "r": 208.421630859375, "t": 643.2987060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "GNMT + RL [38]" }, { "bbox": { "b": 634.3921508789062, "coord_origin": "BOTTOMLEFT", "l": 294.93682861328125, "r": 312.3713684082031, "t": 643.2987060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "24.6" }, { "bbox": { "b": 634.3921508789062, "coord_origin": "BOTTOMLEFT", "l": 333.7311706542969, "r": 356.14703369140625, "t": 643.2987060546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "39.92" }, { "bbox": { "b": 634.611328125, "coord_origin": "BOTTOMLEFT", "l": 383.2449951171875, "r": 421.0761413574219, "t": 644.9989013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "2 . 3 · 10 19" }, { "bbox": { "b": 634.611328125, "coord_origin": "BOTTOMLEFT", "l": 435.2640075683594, "r": 473.09515380859375, "t": 644.9989013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "1 . 4 · 10 20" }, { "bbox": { "b": 623.0091552734375, "coord_origin": "BOTTOMLEFT", "l": 136.6710205078125, "r": 188.02822875976562, "t": 631.9157104492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "ConvS2S [9]" }, { "bbox": { "b": 623.0091552734375, "coord_origin": "BOTTOMLEFT", "l": 292.44622802734375, "r": 314.8620910644531, "t": 631.9157104492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "25.16" }, { "bbox": { "b": 623.0091552734375, "coord_origin": "BOTTOMLEFT", "l": 333.73126220703125, "r": 356.1471252441406, "t": 631.9157104492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "40.46" }, { "bbox": { "b": 623.2283325195312, "coord_origin": "BOTTOMLEFT", "l": 383.2450256347656, "r": 421.076171875, "t": 633.616943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 5, "text": "9 . 6 · 10 18" }, { "bbox": { "b": 623.2283325195312, "coord_origin": "BOTTOMLEFT", "l": 435.2640380859375, "r": 473.0951843261719, "t": 633.616943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 5, "text": "1 . 5 · 10 20" }, { "bbox": { "b": 611.627197265625, "coord_origin": "BOTTOMLEFT", "l": 136.67105102539062, "r": 175.68458557128906, "t": 620.5337524414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "MoE [32]" }, { "bbox": { "b": 611.627197265625, "coord_origin": "BOTTOMLEFT", "l": 292.4462585449219, "r": 314.86212158203125, "t": 620.5337524414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "26.03" }, { "bbox": { "b": 611.627197265625, "coord_origin": "BOTTOMLEFT", "l": 333.7312927246094, "r": 356.14715576171875, "t": 620.5337524414062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "40.56" }, { "bbox": { "b": 611.8463745117188, "coord_origin": "BOTTOMLEFT", "l": 383.24505615234375, "r": 421.0762023925781, "t": 622.2339477539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 6, "text": "2 . 0 · 10 19" }, { "bbox": { "b": 611.8463745117188, "coord_origin": "BOTTOMLEFT", "l": 435.2650451660156, "r": 473.0951843261719, "t": 622.2339477539062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 6, "text": "1 . 2 · 10 20" }, { "bbox": { "b": 598.9890747070312, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 276.76507568359375, "t": 607.8956298828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Deep-Att + PosUnk Ensemble [39]" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 598.9890747070312, "coord_origin": "BOTTOMLEFT", "l": 336.22186279296875, "r": 353.6564025878906, "t": 607.8956298828125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "40.4" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 599.208251953125, "coord_origin": "BOTTOMLEFT", "l": 435.26397705078125, "r": 473.0951232910156, "t": 609.5968627929688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 7, "text": "8 . 0 · 10 20" }, { "bbox": { "b": 587.6071166992188, "coord_origin": "BOTTOMLEFT", "l": 136.67098999023438, "r": 250.20477294921875, "t": 596.513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "GNMT + RL Ensemble [38]" }, { "bbox": { "b": 587.6071166992188, "coord_origin": "BOTTOMLEFT", "l": 292.4461975097656, "r": 314.862060546875, "t": 596.513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "26.30" }, { "bbox": { "b": 587.6071166992188, "coord_origin": "BOTTOMLEFT", "l": 333.7312316894531, "r": 356.1470947265625, "t": 596.513671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "41.16" }, { "bbox": { "b": 587.8262939453125, "coord_origin": "BOTTOMLEFT", "l": 383.2449951171875, "r": 421.0761413574219, "t": 598.2138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 8, "text": "1 . 8 · 10 20" }, { "bbox": { "b": 587.8262939453125, "coord_origin": "BOTTOMLEFT", "l": 435.2640075683594, "r": 473.09515380859375, "t": 598.2138671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 8, "text": "1 . 1 · 10 21" }, { "bbox": { "b": 576.22509765625, "coord_origin": "BOTTOMLEFT", "l": 136.6710205078125, "r": 229.81137084960938, "t": 585.1316528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "ConvS2S Ensemble [9]" }, { "bbox": { "b": 576.22509765625, "coord_origin": "BOTTOMLEFT", "l": 292.44622802734375, "r": 314.8620910644531, "t": 585.1316528320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "26.36" }, { "bbox": { "b": 576.2948608398438, "coord_origin": "BOTTOMLEFT", "l": 333.73602294921875, "r": 356.1518859863281, "t": 585.251220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "41.29" }, { "bbox": { "b": 576.4442749023438, "coord_origin": "BOTTOMLEFT", "l": 383.2450256347656, "r": 421.076171875, "t": 586.8318481445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 9, "text": "7 . 7 · 10 19" }, { "bbox": { "b": 576.4442749023438, "coord_origin": "BOTTOMLEFT", "l": 435.2640380859375, "r": 473.0951843261719, "t": 586.8318481445312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 9, "text": "1 . 2 · 10 21" }, { "bbox": { "b": 563.089111328125, "coord_origin": "BOTTOMLEFT", "l": 136.67100524902344, "r": 240.34181213378906, "t": 571.9956665039062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Transformer (base model)" }, { "bbox": { "b": 563.089111328125, "coord_origin": "BOTTOMLEFT", "l": 294.9368591308594, "r": 312.37139892578125, "t": 571.9956665039062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "27.3" }, { "bbox": { "b": 563.089111328125, "coord_origin": "BOTTOMLEFT", "l": 336.22186279296875, "r": 353.6564025878906, "t": 571.9956665039062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "38.1" }, { "bbox": { "b": 563.3082885742188, "coord_origin": "BOTTOMLEFT", "l": 407.3389892578125, "r": 427.7060546875, "t": 572.7129516601562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 10, "text": "3 . 3 ·" }, { "bbox": { "b": 563.3082885742188, "coord_origin": "BOTTOMLEFT", "l": 430.25299072265625, "r": 450.73687744140625, "t": 573.6958618164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 10, "text": "10 18" }, { "bbox": { "b": 551.7070922851562, "coord_origin": "BOTTOMLEFT", "l": 136.67098999023438, "r": 207.97329711914062, "t": 560.6136474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": true, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Transformer (big)" }, { "bbox": { "b": 551.77685546875, "coord_origin": "BOTTOMLEFT", "l": 294.94097900390625, "r": 312.3755187988281, "t": 560.7332153320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "28.4" }, { "bbox": { "b": 551.77685546875, "coord_origin": "BOTTOMLEFT", "l": 336.2259826660156, "r": 353.6605224609375, "t": 560.7332153320312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "41.8" }, { "bbox": { "b": 551.92626953125, "coord_origin": "BOTTOMLEFT", "l": 410.12298583984375, "r": 440.01055908203125, "t": 561.3309326171875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 11, "text": "2 . 3 · 10" }, { "bbox": { "b": 556.12109375, "coord_origin": "BOTTOMLEFT", "l": 440.010986328125, "r": 447.9541320800781, "t": 562.3138427734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 11, "text": "19" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 547.9702758789062, "coord_origin": "BOTTOMLEFT", "l": 130.7423553466797, "r": 482.045654296875, "t": 696.5935668945312 }, "charspan": [ 0, 0 ], "page_no": 8 } ], "references": [], "self_ref": "#/tables/1" }, { "captions": [ { "$ref": "#/texts/111" } ], "children": [], "data": { "grid": [ [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 146.1269989013672, "r": 154.13194274902344, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "N" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 167.17298889160156, "r": 189.79249572753906, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "d$_{model}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 207.1320037841797, "r": 216.78721618652344, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "d$_{ff}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 236.23800659179688, "r": 241.97845458984375, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "h" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 258.4765319824219, "r": 267.8932189941406, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "d$_{k}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 285.4560241699219, "r": 294.6258544921875, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "d$_{v}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 309.843017578125, "r": 332.3408203125, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 8, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 0, "text": "P$_{drop}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 345.5880126953125, "r": 355.9537658691406, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 9, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 0, "text": "ϵ$_{ls}$" }, { "bbox": { "b": 639.4281005859375, "coord_origin": "BOTTOMLEFT", "l": 370.3070068359375, "r": 390.2322082519531, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 10, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 0, "text": "train steps" }, { "bbox": { "b": 639.4281005859375, "coord_origin": "BOTTOMLEFT", "l": 403.2929992675781, "r": 424.0650329589844, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 11, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 0, "text": "PPL (dev)" }, { "bbox": { "b": 639.4281005859375, "coord_origin": "BOTTOMLEFT", "l": 436.0199279785156, "r": 462.03228759765625, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 12, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 0, "text": "BLEU (dev)" }, { "bbox": { "b": 639.6472778320312, "coord_origin": "BOTTOMLEFT", "l": 473.9873962402344, "r": 502.7593994140625, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 13, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 0, "text": "params × 10 6" } ], [ { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 116.46800231933594, "r": 134.17153930664062, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "base" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "6" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 171.25538635253906, "r": 186.19927978515625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "512" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 202.2490234375, "r": 222.17422485351562, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "2048" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 236.6199951171875, "r": 241.60128784179688, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "8" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 258.5377197265625, "r": 268.50030517578125, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "64" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 285.4367370605469, "r": 295.3993225097656, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "64" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 315.11529541015625, "r": 327.56854248046875, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 1, "text": "0.1" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 344.79388427734375, "r": 357.24713134765625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 1, "text": "0.1" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 369.2022399902344, "r": 391.3391418457031, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 1, "text": "100K" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 1, "text": "4.92" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 1, "text": "25.8" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 1, "text": "65" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 236.61700439453125, "r": 241.59829711914062, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "1" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 256.0440673828125, "r": 270.98797607421875, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 2, "text": "512" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 282.9430847167969, "r": 297.8869934082031, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 2, "text": "512" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 2, "text": "" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 2, "text": "5.29" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 2, "text": "24.9" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 2, "text": "" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 236.61700439453125, "r": 241.59829711914062, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "4" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 256.0440673828125, "r": 270.98797607421875, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 3, "text": "128" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 282.9430847167969, "r": 297.8869934082031, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 3, "text": "128" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 3, "text": "5.00" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 3, "text": "25.5" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 3, "text": "" } ], [ { "bbox": { "b": 597.7890625, "coord_origin": "BOTTOMLEFT", "l": 118.40599822998047, "r": 132.2340850830078, "t": 606.6956176757812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "(A)" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 234.1269989013672, "r": 244.089599609375, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "16" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 258.5353698730469, "r": 268.4979553222656, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 4, "text": "32" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 285.43438720703125, "r": 295.39697265625, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 4, "text": "32" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 4, "text": "" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 4, "text": "4.91" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 4, "text": "25.8" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 4, "text": "" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 5, "text": "" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 234.1269989013672, "r": 244.089599609375, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 5, "text": "32" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 258.5353698730469, "r": 268.4979553222656, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 5, "text": "16" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 285.43438720703125, "r": 295.39697265625, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 5, "text": "16" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 5, "text": "" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 5, "text": "5.01" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 5, "text": "25.4" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 5, "text": "" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 6, "text": "" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 258.5350036621094, "r": 268.4975891113281, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 6, "text": "16" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 6, "text": "" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 6, "text": "5.16" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 6, "text": "25.1" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 6, "text": "58" } ], [ { "bbox": { "b": 563.3330688476562, "coord_origin": "BOTTOMLEFT", "l": 118.68000030517578, "r": 131.96014404296875, "t": 572.2396240234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "(B)" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 258.5350036621094, "r": 268.4975891113281, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 7, "text": "32" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 7, "text": "5.01" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 7, "text": "25.4" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 7, "text": "60" } ], [ { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 118.68000030517578, "r": 131.96014404296875, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "(C)" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "2" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 8, "text": "" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 8, "text": "6.11" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 8, "text": "23.7" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 8, "text": "36" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "4" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 9, "text": "" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 9, "text": "5.19" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 9, "text": "25.3" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 9, "text": "50" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "8" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 10, "text": "" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 10, "text": "4.88" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 10, "text": "25.5" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 10, "text": "80" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 171.25999450683594, "r": 186.20388793945312, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "256" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 258.5323486328125, "r": 268.49493408203125, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 11, "text": "32" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 285.4313659667969, "r": 295.3939514160156, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 11, "text": "32" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 11, "text": "5.75" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 11, "text": "24.5" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 11, "text": "28" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 168.7689971923828, "r": 188.69419860839844, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "1024" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 12, "text": "" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 256.0413818359375, "r": 270.98529052734375, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 12, "text": "128" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 282.9403991699219, "r": 297.8843078613281, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 12, "text": "128" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 12, "text": "" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 12, "text": "4.66" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 12, "text": "26.0" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 480.89691162109375, "r": 495.8408203125, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 12, "text": "168" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 202.24600219726562, "r": 222.17120361328125, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "1024" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 13, "text": "" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 13, "text": "5.12" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 13, "text": "25.4" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 13, "text": "53" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 202.24600219726562, "r": 222.17120361328125, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "4096" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 14, "text": "4.75" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 14, "text": "26.2" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 14, "text": "90" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 15, "text": "" }, { "bbox": { "b": 467.1490783691406, "coord_origin": "BOTTOMLEFT", "l": 315.1130065917969, "r": 327.5662536621094, "t": 476.0556335449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 15, "text": "0.0" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 15, "text": "" }, { "bbox": { "b": 467.1490783691406, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 476.0556335449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 15, "text": "5.77" }, { "bbox": { "b": 467.1490783691406, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 476.0556335449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 15, "text": "24.6" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 15, "text": "" } ], [ { "bbox": { "b": 450.78509521484375, "coord_origin": "BOTTOMLEFT", "l": 118.40599822998047, "r": 132.2340850830078, "t": 459.691650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "(D)" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 16, "text": "" }, { "bbox": { "b": 456.2400817871094, "coord_origin": "BOTTOMLEFT", "l": 315.1130065917969, "r": 327.5662536621094, "t": 465.1466369628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 16, "text": "0.2" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 16, "text": "" }, { "bbox": { "b": 456.2400817871094, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 465.1466369628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 16, "text": "4.95" }, { "bbox": { "b": 456.2400817871094, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 465.1466369628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 16, "text": "25.5" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 16, "text": "" } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 17, "text": "" }, { "bbox": { "b": 445.3310852050781, "coord_origin": "BOTTOMLEFT", "l": 344.7919921875, "r": 357.2452392578125, "t": 454.2376403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 17, "text": "0.0" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 17, "text": "" }, { "bbox": { "b": 445.3310852050781, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 454.2376403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 17, "text": "4.67" }, { "bbox": { "b": 445.3310852050781, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 454.2376403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 17, "text": "25.3" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 17, "text": "" } ], [ { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 118.95899963378906, "r": 131.68124389648438, "t": 430.6906433105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "(E)" }, { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 178.63400268554688, "r": 357.2452392578125, "t": 443.3286437988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "0.2 positional embedding instead of sinusoids" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 18, "text": "" }, { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 443.3286437988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 18, "text": "5.47 4.92" }, { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 443.3286437988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 18, "text": "25.7 25.7" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 18, "text": "" } ], [ { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 118.9540023803711, "r": 131.6862030029297, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "big" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "6" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 168.7647247314453, "r": 188.68992614746094, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "1024" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 202.2490234375, "r": 222.17422485351562, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 19, "text": "4096" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 234.1293487548828, "r": 244.09194946289062, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 19, "text": "16" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 19, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 19, "text": "" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 315.1153259277344, "r": 327.5685729980469, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 19, "text": "0.3" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 19, "text": "" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 369.2022705078125, "r": 391.33917236328125, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 19, "text": "300K" }, { "bbox": { "b": 409.2168273925781, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 418.1732177734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 19, "text": "4.33" }, { "bbox": { "b": 409.2168273925781, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 418.1732177734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 19, "text": "26.4" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 480.9020080566406, "r": 495.8459167480469, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 19, "text": "213" } ] ], "num_cols": 13, "num_rows": 20, "table_cells": [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 146.1269989013672, "r": 154.13194274902344, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "N" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 167.17298889160156, "r": 189.79249572753906, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "d$_{model}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 207.1320037841797, "r": 216.78721618652344, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "d$_{ff}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 236.23800659179688, "r": 241.97845458984375, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "h" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 258.4765319824219, "r": 267.8932189941406, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 6, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 0, "text": "d$_{k}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 285.4560241699219, "r": 294.6258544921875, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 7, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 0, "text": "d$_{v}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 309.843017578125, "r": 332.3408203125, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 8, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 0, "text": "P$_{drop}$" }, { "bbox": { "b": 645.5752563476562, "coord_origin": "BOTTOMLEFT", "l": 345.5880126953125, "r": 355.9537658691406, "t": 654.4220581054688 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 9, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 0, "text": "ϵ$_{ls}$" }, { "bbox": { "b": 639.4281005859375, "coord_origin": "BOTTOMLEFT", "l": 370.3070068359375, "r": 390.2322082519531, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 10, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 0, "text": "train steps" }, { "bbox": { "b": 639.4281005859375, "coord_origin": "BOTTOMLEFT", "l": 403.2929992675781, "r": 424.0650329589844, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 11, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 0, "text": "PPL (dev)" }, { "bbox": { "b": 639.4281005859375, "coord_origin": "BOTTOMLEFT", "l": 436.0199279785156, "r": 462.03228759765625, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 12, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 0, "text": "BLEU (dev)" }, { "bbox": { "b": 639.6472778320312, "coord_origin": "BOTTOMLEFT", "l": 473.9873962402344, "r": 502.7593994140625, "t": 659.7166137695312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 13, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 0, "text": "params × 10 6" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 116.46800231933594, "r": 134.17153930664062, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "base" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "6" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 171.25538635253906, "r": 186.19927978515625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "512" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 202.2490234375, "r": 222.17422485351562, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "2048" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 236.6199951171875, "r": 241.60128784179688, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "8" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 258.5377197265625, "r": 268.50030517578125, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 1, "text": "64" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 285.4367370605469, "r": 295.3993225097656, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 1, "text": "64" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 315.11529541015625, "r": 327.56854248046875, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 1, "text": "0.1" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 344.79388427734375, "r": 357.24713134765625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 1, "text": "0.1" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 369.2022399902344, "r": 391.3391418457031, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 1, "text": "100K" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 1, "text": "4.92" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 1, "text": "25.8" }, { "bbox": { "b": 626.7901000976562, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 635.6966552734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 1, "text": "65" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 236.61700439453125, "r": 241.59829711914062, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "1" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 256.0440673828125, "r": 270.98797607421875, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 2, "text": "512" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 282.9430847167969, "r": 297.8869934082031, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 2, "text": "512" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 2, "text": "" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 2, "text": "5.29" }, { "bbox": { "b": 614.153076171875, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 623.0596313476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 2, "text": "24.9" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 2, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 236.61700439453125, "r": 241.59829711914062, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "4" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 256.0440673828125, "r": 270.98797607421875, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 3, "text": "128" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 282.9430847167969, "r": 297.8869934082031, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 3, "text": "128" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 3, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 3, "text": "5.00" }, { "bbox": { "b": 603.2431030273438, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 612.149658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 3, "text": "25.5" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 3, "text": "" }, { "bbox": { "b": 597.7890625, "coord_origin": "BOTTOMLEFT", "l": 118.40599822998047, "r": 132.2340850830078, "t": 606.6956176757812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "(A)" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 234.1269989013672, "r": 244.089599609375, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "16" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 258.5353698730469, "r": 268.4979553222656, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 4, "text": "32" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 285.43438720703125, "r": 295.39697265625, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 4, "text": "32" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 4, "text": "" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 4, "text": "4.91" }, { "bbox": { "b": 592.3341064453125, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 601.2406616210938 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 4, "text": "25.8" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 4, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 5, "text": "" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 234.1269989013672, "r": 244.089599609375, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 5, "text": "32" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 258.5353698730469, "r": 268.4979553222656, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 5, "text": "16" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 285.43438720703125, "r": 295.39697265625, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 5, "text": "16" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 5, "text": "" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 5, "text": "5.01" }, { "bbox": { "b": 581.4251098632812, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 590.3316650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 5, "text": "25.4" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 5, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 6, "text": "" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 258.5350036621094, "r": 268.4975891113281, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 6, "text": "16" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 6, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 6, "text": "" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 6, "text": "5.16" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 6, "text": "25.1" }, { "bbox": { "b": 568.7880859375, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 577.6946411132812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 6, "text": "58" }, { "bbox": { "b": 563.3330688476562, "coord_origin": "BOTTOMLEFT", "l": 118.68000030517578, "r": 131.96014404296875, "t": 572.2396240234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "(B)" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 258.5350036621094, "r": 268.4975891113281, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 7, "text": "32" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 7, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 7, "text": "" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 7, "text": "5.01" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 7, "text": "25.4" }, { "bbox": { "b": 557.8790893554688, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 566.78564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 7, "text": "60" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 118.68000030517578, "r": 131.96014404296875, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "(C)" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "2" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 8, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 8, "text": "" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 8, "text": "6.11" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 8, "text": "23.7" }, { "bbox": { "b": 545.2410888671875, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 554.1476440429688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 8, "text": "36" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "4" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 9, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 9, "text": "" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 9, "text": "5.19" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 9, "text": "25.3" }, { "bbox": { "b": 534.3320922851562, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 543.2386474609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 9, "text": "50" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "8" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 10, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 10, "text": "" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 10, "text": "4.88" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 10, "text": "25.5" }, { "bbox": { "b": 523.423095703125, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 532.3296508789062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 10, "text": "80" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 171.25999450683594, "r": 186.20388793945312, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "256" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 258.5323486328125, "r": 268.49493408203125, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 11, "text": "32" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 285.4313659667969, "r": 295.3939514160156, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 11, "text": "32" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 11, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 11, "text": "" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 11, "text": "5.75" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 11, "text": "24.5" }, { "bbox": { "b": 512.5140991210938, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 521.420654296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 11, "text": "28" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 168.7689971923828, "r": 188.69419860839844, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "1024" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 12, "text": "" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 256.0413818359375, "r": 270.98529052734375, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 12, "text": "128" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 282.9403991699219, "r": 297.8843078613281, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 12, "text": "128" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 12, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 12, "text": "" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 12, "text": "4.66" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 12, "text": "26.0" }, { "bbox": { "b": 501.6050720214844, "coord_origin": "BOTTOMLEFT", "l": 480.89691162109375, "r": 495.8408203125, "t": 510.5116271972656 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 12, "text": "168" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 202.24600219726562, "r": 222.17120361328125, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 13, "text": "1024" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 13, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 13, "text": "" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 13, "text": "5.12" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 13, "text": "25.4" }, { "bbox": { "b": 490.6960754394531, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 499.6026306152344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 13, "text": "53" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 202.24600219726562, "r": 222.17120361328125, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "4096" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 14, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 14, "text": "" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 14, "text": "4.75" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 14, "text": "26.2" }, { "bbox": { "b": 479.7870788574219, "coord_origin": "BOTTOMLEFT", "l": 483.3875732421875, "r": 493.35015869140625, "t": 488.6936340332031 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 14, "text": "90" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 15, "text": "" }, { "bbox": { "b": 467.1490783691406, "coord_origin": "BOTTOMLEFT", "l": 315.1130065917969, "r": 327.5662536621094, "t": 476.0556335449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 15, "text": "0.0" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 15, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 15, "text": "" }, { "bbox": { "b": 467.1490783691406, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 476.0556335449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 15, "text": "5.77" }, { "bbox": { "b": 467.1490783691406, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 476.0556335449219 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 15, "text": "24.6" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 15, "text": "" }, { "bbox": { "b": 450.78509521484375, "coord_origin": "BOTTOMLEFT", "l": 118.40599822998047, "r": 132.2340850830078, "t": 459.691650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "(D)" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 16, "text": "" }, { "bbox": { "b": 456.2400817871094, "coord_origin": "BOTTOMLEFT", "l": 315.1130065917969, "r": 327.5662536621094, "t": 465.1466369628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 16, "text": "0.2" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 16, "text": "" }, { "bbox": { "b": 456.2400817871094, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 465.1466369628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 16, "text": "4.95" }, { "bbox": { "b": 456.2400817871094, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 465.1466369628906 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 16, "text": "25.5" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 16, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 17, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 17, "text": "" }, { "bbox": { "b": 445.3310852050781, "coord_origin": "BOTTOMLEFT", "l": 344.7919921875, "r": 357.2452392578125, "t": 454.2376403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 17, "text": "0.0" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 17, "text": "" }, { "bbox": { "b": 445.3310852050781, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 454.2376403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 17, "text": "4.67" }, { "bbox": { "b": 445.3310852050781, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 454.2376403808594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 17, "text": "25.3" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 17, "text": "" }, { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 118.95899963378906, "r": 131.68124389648438, "t": 430.6906433105469 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "(E)" }, { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 178.63400268554688, "r": 357.2452392578125, "t": 443.3286437988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "0.2 positional embedding instead of sinusoids" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 18, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 18, "text": "" }, { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 443.3286437988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 18, "text": "5.47 4.92" }, { "bbox": { "b": 421.7840881347656, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 443.3286437988281 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 18, "text": "25.7 25.7" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 18, "text": "" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 118.9540023803711, "r": 131.6862030029297, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "big" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 148.1820068359375, "r": 153.16329956054688, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "6" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 168.7647247314453, "r": 188.68992614746094, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 19, "text": "1024" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 202.2490234375, "r": 222.17422485351562, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 19, "text": "4096" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 234.1293487548828, "r": 244.09194946289062, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 19, "text": "16" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 6, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 5, "start_row_offset_idx": 19, "text": "" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 7, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 6, "start_row_offset_idx": 19, "text": "" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 315.1153259277344, "r": 327.5685729980469, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 8, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 7, "start_row_offset_idx": 19, "text": "0.3" }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 9, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 8, "start_row_offset_idx": 19, "text": "" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 369.2022705078125, "r": 391.33917236328125, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 10, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 9, "start_row_offset_idx": 19, "text": "300K" }, { "bbox": { "b": 409.2168273925781, "coord_origin": "BOTTOMLEFT", "l": 404.9620056152344, "r": 422.39654541015625, "t": 418.1732177734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 11, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 10, "start_row_offset_idx": 19, "text": "4.33" }, { "bbox": { "b": 409.2168273925781, "coord_origin": "BOTTOMLEFT", "l": 440.3092956542969, "r": 457.74383544921875, "t": 418.1732177734375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 12, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 11, "start_row_offset_idx": 19, "text": "26.4" }, { "bbox": { "b": 409.1470947265625, "coord_origin": "BOTTOMLEFT", "l": 480.9020080566406, "r": 495.8459167480469, "t": 418.05364990234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 13, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 12, "start_row_offset_idx": 19, "text": "213" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 407.0993347167969, "coord_origin": "BOTTOMLEFT", "l": 107.06884002685547, "r": 509.0139465332031, "t": 662.7780151367188 }, "charspan": [ 0, 0 ], "page_no": 9 } ], "references": [], "self_ref": "#/tables/2" }, { "captions": [ { "$ref": "#/texts/120" } ], "children": [], "data": { "grid": [ [ { "bbox": { "b": 688.1348266601562, "coord_origin": "BOTTOMLEFT", "l": 206.75799560546875, "r": 234.87245178222656, "t": 697.0911865234375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Parser" }, { "bbox": { "b": 688.1348266601562, "coord_origin": "BOTTOMLEFT", "l": 334.0840148925781, "r": 370.99542236328125, "t": 697.0911865234375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Training" }, { "bbox": { "b": 688.1348266601562, "coord_origin": "BOTTOMLEFT", "l": 414.47698974609375, "r": 460.9724426269531, "t": 697.0911865234375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "WSJ 23 F1" } ], [ { "bbox": { "b": 676.758056640625, "coord_origin": "BOTTOMLEFT", "l": 151.02699279785156, "r": 290.6029968261719, "t": 685.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Vinyals & Kaiser el al. (2014) [37]" }, { "bbox": { "b": 676.758056640625, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 685.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "WSJ only, discriminative" }, { "bbox": { "b": 676.758056640625, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 685.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "88.3" } ], [ { "bbox": { "b": 665.8490600585938, "coord_origin": "BOTTOMLEFT", "l": 172.58599853515625, "r": 269.0438537597656, "t": 674.755615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Petrov et al. (2006) [29]" }, { "bbox": { "b": 665.8490600585938, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 674.755615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "WSJ only, discriminative" }, { "bbox": { "b": 665.8490600585938, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 674.755615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "90.4" } ], [ { "bbox": { "b": 654.9400634765625, "coord_origin": "BOTTOMLEFT", "l": 177.4929962158203, "r": 264.1376953125, "t": 663.8466186523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Zhu et al. (2013) [40]" }, { "bbox": { "b": 654.9400634765625, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 663.8466186523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "WSJ only, discriminative" }, { "bbox": { "b": 654.9400634765625, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 663.8466186523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "90.4" } ], [ { "bbox": { "b": 644.0310668945312, "coord_origin": "BOTTOMLEFT", "l": 178.05099487304688, "r": 263.57989501953125, "t": 652.9376220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Dyer et al. (2016) [8]" }, { "bbox": { "b": 644.0310668945312, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 652.9376220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "WSJ only, discriminative" }, { "bbox": { "b": 644.0310668945312, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 652.9376220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "91.7" } ], [ { "bbox": { "b": 633.1220703125, "coord_origin": "BOTTOMLEFT", "l": 175.8990020751953, "r": 265.73175048828125, "t": 642.0286254882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Transformer (4 layers)" }, { "bbox": { "b": 633.1220703125, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 642.0286254882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "WSJ only, discriminative" }, { "bbox": { "b": 633.1220703125, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 642.0286254882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "91.3" } ], [ { "bbox": { "b": 622.2130737304688, "coord_origin": "BOTTOMLEFT", "l": 177.4929962158203, "r": 264.1376953125, "t": 631.11962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Zhu et al. (2013) [40]" }, { "bbox": { "b": 622.2130737304688, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 631.11962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "semi-supervised" }, { "bbox": { "b": 622.2130737304688, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 631.11962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "91.3" } ], [ { "bbox": { "b": 611.3031005859375, "coord_origin": "BOTTOMLEFT", "l": 163.27099609375, "r": 278.35894775390625, "t": 620.2096557617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Huang & Harper (2009) [14]" }, { "bbox": { "b": 611.3031005859375, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 620.2096557617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "semi-supervised" }, { "bbox": { "b": 611.3031005859375, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 620.2096557617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "91.3" } ], [ { "bbox": { "b": 600.3941040039062, "coord_origin": "BOTTOMLEFT", "l": 164.83499145507812, "r": 276.794677734375, "t": 609.3006591796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "McClosky et al. (2006) [26]" }, { "bbox": { "b": 600.3941040039062, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 609.3006591796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "semi-supervised" }, { "bbox": { "b": 600.3941040039062, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 609.3006591796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "92.1" } ], [ { "bbox": { "b": 589.485107421875, "coord_origin": "BOTTOMLEFT", "l": 151.02700805664062, "r": 290.6029968261719, "t": 598.3916625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Vinyals & Kaiser el al. (2014) [37]" }, { "bbox": { "b": 589.485107421875, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 598.3916625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "semi-supervised" }, { "bbox": { "b": 589.485107421875, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 598.3916625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "92.1" } ], [ { "bbox": { "b": 578.5761108398438, "coord_origin": "BOTTOMLEFT", "l": 175.8990020751953, "r": 265.73175048828125, "t": 587.482666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Transformer (4 layers)" }, { "bbox": { "b": 578.5761108398438, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 587.482666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "semi-supervised" }, { "bbox": { "b": 578.5761108398438, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 587.482666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "92.7" } ], [ { "bbox": { "b": 567.6670532226562, "coord_origin": "BOTTOMLEFT", "l": 172.51100158691406, "r": 269.1183166503906, "t": 576.5736083984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Luong et al. (2015) [23]" }, { "bbox": { "b": 567.6670532226562, "coord_origin": "BOTTOMLEFT", "l": 332.33599853515625, "r": 372.7442932128906, "t": 576.5736083984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "multi-task" }, { "bbox": { "b": 567.6670532226562, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 576.5736083984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "93.0" } ], [ { "bbox": { "b": 556.758056640625, "coord_origin": "BOTTOMLEFT", "l": 178.05099487304688, "r": 263.57989501953125, "t": 565.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Dyer et al. (2016) [8]" }, { "bbox": { "b": 556.758056640625, "coord_origin": "BOTTOMLEFT", "l": 331.99200439453125, "r": 373.0877380371094, "t": 565.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "generative" }, { "bbox": { "b": 556.758056640625, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 565.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "93.3" } ] ], "num_cols": 3, "num_rows": 13, "table_cells": [ { "bbox": { "b": 688.1348266601562, "coord_origin": "BOTTOMLEFT", "l": 206.75799560546875, "r": 234.87245178222656, "t": 697.0911865234375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Parser" }, { "bbox": { "b": 688.1348266601562, "coord_origin": "BOTTOMLEFT", "l": 334.0840148925781, "r": 370.99542236328125, "t": 697.0911865234375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Training" }, { "bbox": { "b": 688.1348266601562, "coord_origin": "BOTTOMLEFT", "l": 414.47698974609375, "r": 460.9724426269531, "t": 697.0911865234375 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "WSJ 23 F1" }, { "bbox": { "b": 676.758056640625, "coord_origin": "BOTTOMLEFT", "l": 151.02699279785156, "r": 290.6029968261719, "t": 685.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Vinyals & Kaiser el al. (2014) [37]" }, { "bbox": { "b": 676.758056640625, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 685.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "WSJ only, discriminative" }, { "bbox": { "b": 676.758056640625, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 685.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "88.3" }, { "bbox": { "b": 665.8490600585938, "coord_origin": "BOTTOMLEFT", "l": 172.58599853515625, "r": 269.0438537597656, "t": 674.755615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Petrov et al. (2006) [29]" }, { "bbox": { "b": 665.8490600585938, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 674.755615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "WSJ only, discriminative" }, { "bbox": { "b": 665.8490600585938, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 674.755615234375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "90.4" }, { "bbox": { "b": 654.9400634765625, "coord_origin": "BOTTOMLEFT", "l": 177.4929962158203, "r": 264.1376953125, "t": 663.8466186523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Zhu et al. (2013) [40]" }, { "bbox": { "b": 654.9400634765625, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 663.8466186523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "WSJ only, discriminative" }, { "bbox": { "b": 654.9400634765625, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 663.8466186523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "90.4" }, { "bbox": { "b": 644.0310668945312, "coord_origin": "BOTTOMLEFT", "l": 178.05099487304688, "r": 263.57989501953125, "t": 652.9376220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Dyer et al. (2016) [8]" }, { "bbox": { "b": 644.0310668945312, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 652.9376220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "WSJ only, discriminative" }, { "bbox": { "b": 644.0310668945312, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 652.9376220703125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "91.7" }, { "bbox": { "b": 633.1220703125, "coord_origin": "BOTTOMLEFT", "l": 175.8990020751953, "r": 265.73175048828125, "t": 642.0286254882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Transformer (4 layers)" }, { "bbox": { "b": 633.1220703125, "coord_origin": "BOTTOMLEFT", "l": 302.5580139160156, "r": 402.5227966308594, "t": 642.0286254882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "WSJ only, discriminative" }, { "bbox": { "b": 633.1220703125, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 642.0286254882812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 5, "text": "91.3" }, { "bbox": { "b": 622.2130737304688, "coord_origin": "BOTTOMLEFT", "l": 177.4929962158203, "r": 264.1376953125, "t": 631.11962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Zhu et al. (2013) [40]" }, { "bbox": { "b": 622.2130737304688, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 631.11962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "semi-supervised" }, { "bbox": { "b": 622.2130737304688, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 631.11962890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 6, "text": "91.3" }, { "bbox": { "b": 611.3031005859375, "coord_origin": "BOTTOMLEFT", "l": 163.27099609375, "r": 278.35894775390625, "t": 620.2096557617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Huang & Harper (2009) [14]" }, { "bbox": { "b": 611.3031005859375, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 620.2096557617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "semi-supervised" }, { "bbox": { "b": 611.3031005859375, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 620.2096557617188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 7, "text": "91.3" }, { "bbox": { "b": 600.3941040039062, "coord_origin": "BOTTOMLEFT", "l": 164.83499145507812, "r": 276.794677734375, "t": 609.3006591796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "McClosky et al. (2006) [26]" }, { "bbox": { "b": 600.3941040039062, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 609.3006591796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "semi-supervised" }, { "bbox": { "b": 600.3941040039062, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 609.3006591796875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 8, "text": "92.1" }, { "bbox": { "b": 589.485107421875, "coord_origin": "BOTTOMLEFT", "l": 151.02700805664062, "r": 290.6029968261719, "t": 598.3916625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "Vinyals & Kaiser el al. (2014) [37]" }, { "bbox": { "b": 589.485107421875, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 598.3916625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "semi-supervised" }, { "bbox": { "b": 589.485107421875, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 598.3916625976562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 9, "text": "92.1" }, { "bbox": { "b": 578.5761108398438, "coord_origin": "BOTTOMLEFT", "l": 175.8990020751953, "r": 265.73175048828125, "t": 587.482666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Transformer (4 layers)" }, { "bbox": { "b": 578.5761108398438, "coord_origin": "BOTTOMLEFT", "l": 320.1669921875, "r": 384.9139404296875, "t": 587.482666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "semi-supervised" }, { "bbox": { "b": 578.5761108398438, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 587.482666015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 10, "text": "92.7" }, { "bbox": { "b": 567.6670532226562, "coord_origin": "BOTTOMLEFT", "l": 172.51100158691406, "r": 269.1183166503906, "t": 576.5736083984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "Luong et al. (2015) [23]" }, { "bbox": { "b": 567.6670532226562, "coord_origin": "BOTTOMLEFT", "l": 332.33599853515625, "r": 372.7442932128906, "t": 576.5736083984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "multi-task" }, { "bbox": { "b": 567.6670532226562, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 576.5736083984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 11, "text": "93.0" }, { "bbox": { "b": 556.758056640625, "coord_origin": "BOTTOMLEFT", "l": 178.05099487304688, "r": 263.57989501953125, "t": 565.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "Dyer et al. (2016) [8]" }, { "bbox": { "b": 556.758056640625, "coord_origin": "BOTTOMLEFT", "l": 331.99200439453125, "r": 373.0877380371094, "t": 565.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "generative" }, { "bbox": { "b": 556.758056640625, "coord_origin": "BOTTOMLEFT", "l": 429.00799560546875, "r": 446.4425354003906, "t": 565.6646118164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 12, "text": "93.3" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 555.0657348632812, "coord_origin": "BOTTOMLEFT", "l": 144.03054809570312, "r": 467.5986633300781, "t": 699.1512451171875 }, "charspan": [ 0, 0 ], "page_no": 10 } ], "references": [], "self_ref": "#/tables/3" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 707.8850708007812, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 124.91486358642578, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "[5]" }, { "bbox": { "b": 686.0670776367188, "coord_origin": "BOTTOMLEFT", "l": 127.47358703613281, "r": 505.24237060546875, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Kyunghyun Cho, Bart van Merrienboer, Caglar Gulcehre, Fethi Bougares, Holger Schwenk, and Yoshua Bengio. Learning phrase representations using rnn encoder-decoder for statistical machine translation. CoRR , abs/1406.1078, 2014." } ], [ { "bbox": { "b": 666.2660522460938, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 125.25128936767578, "t": 675.172607421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "[6]" }, { "bbox": { "b": 655.3570556640625, "coord_origin": "BOTTOMLEFT", "l": 127.88213348388672, "r": 504.0006103515625, "t": 675.172607421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Francois Chollet. Xception: Deep learning with depthwise separable convolutions. arXiv preprint arXiv:1610.02357 , 2016." } ], [ { "bbox": { "b": 635.5560302734375, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 124.46833801269531, "t": 644.4625854492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "[7]" }, { "bbox": { "b": 624.6470947265625, "coord_origin": "BOTTOMLEFT", "l": 126.9312973022461, "r": 504.00372314453125, "t": 644.4625854492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Junyoung Chung, Çaglar Gülçehre, Kyunghyun Cho, and Yoshua Bengio. Empirical evaluation of gated recurrent neural networks on sequence modeling. CoRR , abs/1412.3555, 2014." } ], [ { "bbox": { "b": 604.8460693359375, "coord_origin": "BOTTOMLEFT", "l": 112.98098754882812, "r": 125.13201904296875, "t": 613.7526245117188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "[8]" }, { "bbox": { "b": 593.9370727539062, "coord_origin": "BOTTOMLEFT", "l": 127.73729705810547, "r": 504.0025329589844, "t": 613.7526245117188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "Chris Dyer, Adhiguna Kuncoro, Miguel Ballesteros, and Noah A. Smith. Recurrent neural network grammars. In Proc. of NAACL , 2016." } ], [ { "bbox": { "b": 574.1360473632812, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 124.73145294189453, "t": 583.0426025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "[9]" }, { "bbox": { "b": 563.2271118164062, "coord_origin": "BOTTOMLEFT", "l": 127.25084686279297, "r": 505.6533203125, "t": 583.0426025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "Jonas Gehring, Michael Auli, David Grangier, Denis Yarats, and Yann N. Dauphin. Convolu- tional sequence to sequence learning. arXiv preprint arXiv:1705.03122v2 , 2017." } ], [ { "bbox": { "b": 543.4260864257812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 125.94074249267578, "t": 552.3326416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "[10]" }, { "bbox": { "b": 532.51708984375, "coord_origin": "BOTTOMLEFT", "l": 128.6329345703125, "r": 503.9955139160156, "t": 552.3326416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Alex Graves. Generating sequences with recurrent neural networks. arXiv preprint arXiv:1308.0850 , 2013." } ], [ { "bbox": { "b": 512.716064453125, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.46177673339844, "t": 521.6226196289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "[11]" }, { "bbox": { "b": 490.8970947265625, "coord_origin": "BOTTOMLEFT", "l": 128.0821075439453, "r": 505.6487731933594, "t": 521.6226196289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for im- age recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 770-778, 2016." } ], [ { "bbox": { "b": 471.0960998535156, "coord_origin": "BOTTOMLEFT", "l": 108.00000762939453, "r": 124.84686279296875, "t": 480.0026550292969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "[12]" }, { "bbox": { "b": 460.18707275390625, "coord_origin": "BOTTOMLEFT", "l": 127.3748779296875, "r": 504.001953125, "t": 480.0026550292969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Sepp Hochreiter, Yoshua Bengio, Paolo Frasconi, and Jürgen Schmidhuber. Gradient flow in recurrent nets: the difficulty of learning long-term dependencies, 2001." } ], [ { "bbox": { "b": 440.3860778808594, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.3443603515625, "t": 449.2926330566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "[13]" }, { "bbox": { "b": 429.4770812988281, "coord_origin": "BOTTOMLEFT", "l": 127.94705963134766, "r": 505.2454528808594, "t": 449.2926330566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory. Neural computation , 9(8):1735-1780, 1997." } ], [ { "bbox": { "b": 409.67608642578125, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.0400619506836, "t": 418.5826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "[14]" }, { "bbox": { "b": 387.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 127.59709930419922, "r": 504.0025939941406, "t": 418.5826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "Zhongqiang Huang and Mary Harper. Self-training PCFG grammars with latent annotations across languages. In Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing , pages 832-841. ACL, August 2009." } ], [ { "bbox": { "b": 368.0570983886719, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.76412963867188, "t": 376.9636535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "[15]" }, { "bbox": { "b": 357.1480712890625, "coord_origin": "BOTTOMLEFT", "l": 127.27974700927734, "r": 503.999755859375, "t": 376.9636535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "Rafal Jozefowicz, Oriol Vinyals, Mike Schuster, Noam Shazeer, and Yonghui Wu. Exploring the limits of language modeling. arXiv preprint arXiv:1602.02410 , 2016." } ], [ { "bbox": { "b": 337.3470764160156, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.39153289794922, "t": 346.2536315917969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "[16]" }, { "bbox": { "b": 326.4380798339844, "coord_origin": "BOTTOMLEFT", "l": 126.85124206542969, "r": 503.9952392578125, "t": 346.2536315917969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "Łukasz Kaiser and Samy Bengio. Can active memory replace attention? In Advances in Neural Information Processing Systems, (NIPS) , 2016." } ], [ { "bbox": { "b": 306.6370849609375, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 124.46515655517578, "t": 315.54364013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "[17]" }, { "bbox": { "b": 295.72808837890625, "coord_origin": "BOTTOMLEFT", "l": 126.93594360351562, "r": 504.0040283203125, "t": 315.54364013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "Łukasz Kaiser and Ilya Sutskever. Neural GPUs learn algorithms. In International Conference on Learning Representations (ICLR) , 2016." } ], [ { "bbox": { "b": 275.9270935058594, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.2394790649414, "t": 284.8336486816406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "[18]" }, { "bbox": { "b": 254.10906982421875, "coord_origin": "BOTTOMLEFT", "l": 126.67637634277344, "r": 505.6534729003906, "t": 284.8336486816406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "Nal Kalchbrenner, Lasse Espeholt, Karen Simonyan, Aaron van den Oord, Alex Graves, and Ko- ray Kavukcuoglu. Neural machine translation in linear time. arXiv preprint arXiv:1610.10099v2 , 2017." } ], [ { "bbox": { "b": 234.30807495117188, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 124.46531677246094, "t": 243.21463012695312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "[19]" }, { "bbox": { "b": 223.39907836914062, "coord_origin": "BOTTOMLEFT", "l": 126.93611145019531, "r": 505.747314453125, "t": 243.21463012695312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "Yoon Kim, Carl Denton, Luong Hoang, and Alexander M. Rush. Structured attention networks. In International Conference on Learning Representations , 2017." } ], [ { "bbox": { "b": 203.59808349609375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.6935806274414, "t": 212.504638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "[20]" }, { "bbox": { "b": 203.59707641601562, "coord_origin": "BOTTOMLEFT", "l": 127.19862365722656, "r": 505.7453308105469, "t": 212.504638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "Diederik Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In ICLR , 2015." } ], [ { "bbox": { "b": 183.79608154296875, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.598876953125, "t": 192.70263671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "[21]" }, { "bbox": { "b": 172.8870849609375, "coord_origin": "BOTTOMLEFT", "l": 127.08970642089844, "r": 504.0007019042969, "t": 192.70263671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "Oleksii Kuchaiev and Boris Ginsburg. Factorization tricks for LSTM networks. arXiv preprint arXiv:1703.10722 , 2017." } ], [ { "bbox": { "b": 153.08609008789062, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 126.07655334472656, "t": 161.99264526367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "[22]" }, { "bbox": { "b": 131.26808166503906, "coord_origin": "BOTTOMLEFT", "l": 128.7891082763672, "r": 504.0025939941406, "t": 161.99264526367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "Zhouhan Lin, Minwei Feng, Cicero Nogueira dos Santos, Mo Yu, Bing Xiang, Bowen Zhou, and Yoshua Bengio. A structured self-attentive sentence embedding. arXiv preprint arXiv:1703.03130 , 2017." } ], [ { "bbox": { "b": 111.46707153320312, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 124.56388092041016, "t": 120.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "[23]" }, { "bbox": { "b": 100.55806732177734, "coord_origin": "BOTTOMLEFT", "l": 127.04947662353516, "r": 504.25091552734375, "t": 120.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "Minh-Thang Luong, Quoc V. Le, Ilya Sutskever, Oriol Vinyals, and Lukasz Kaiser. Multi-task sequence to sequence learning. arXiv preprint arXiv:1511.06114 , 2015." } ], [ { "bbox": { "b": 80.75707244873047, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.32675170898438, "t": 89.66363525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "[24]" }, { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 126.77674102783203, "r": 505.6534118652344, "t": 89.66363525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "Minh-Thang Luong, Hieu Pham, and Christopher D Manning. Effective approaches to attention- arXiv preprint arXiv:1508.04025 , 2015." } ], [ { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "" }, { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 129.57899475097656, "r": 262.9383544921875, "t": 78.754638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "based neural machine translation." } ] ], "num_cols": 2, "num_rows": 21, "table_cells": [ { "bbox": { "b": 707.8850708007812, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 124.91486358642578, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "[5]" }, { "bbox": { "b": 686.0670776367188, "coord_origin": "BOTTOMLEFT", "l": 127.47358703613281, "r": 505.24237060546875, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Kyunghyun Cho, Bart van Merrienboer, Caglar Gulcehre, Fethi Bougares, Holger Schwenk, and Yoshua Bengio. Learning phrase representations using rnn encoder-decoder for statistical machine translation. CoRR , abs/1406.1078, 2014." }, { "bbox": { "b": 666.2660522460938, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 125.25128936767578, "t": 675.172607421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "[6]" }, { "bbox": { "b": 655.3570556640625, "coord_origin": "BOTTOMLEFT", "l": 127.88213348388672, "r": 504.0006103515625, "t": 675.172607421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Francois Chollet. Xception: Deep learning with depthwise separable convolutions. arXiv preprint arXiv:1610.02357 , 2016." }, { "bbox": { "b": 635.5560302734375, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 124.46833801269531, "t": 644.4625854492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "[7]" }, { "bbox": { "b": 624.6470947265625, "coord_origin": "BOTTOMLEFT", "l": 126.9312973022461, "r": 504.00372314453125, "t": 644.4625854492188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Junyoung Chung, Çaglar Gülçehre, Kyunghyun Cho, and Yoshua Bengio. Empirical evaluation of gated recurrent neural networks on sequence modeling. CoRR , abs/1412.3555, 2014." }, { "bbox": { "b": 604.8460693359375, "coord_origin": "BOTTOMLEFT", "l": 112.98098754882812, "r": 125.13201904296875, "t": 613.7526245117188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "[8]" }, { "bbox": { "b": 593.9370727539062, "coord_origin": "BOTTOMLEFT", "l": 127.73729705810547, "r": 504.0025329589844, "t": 613.7526245117188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "Chris Dyer, Adhiguna Kuncoro, Miguel Ballesteros, and Noah A. Smith. Recurrent neural network grammars. In Proc. of NAACL , 2016." }, { "bbox": { "b": 574.1360473632812, "coord_origin": "BOTTOMLEFT", "l": 112.98100280761719, "r": 124.73145294189453, "t": 583.0426025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "[9]" }, { "bbox": { "b": 563.2271118164062, "coord_origin": "BOTTOMLEFT", "l": 127.25084686279297, "r": 505.6533203125, "t": 583.0426025390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "Jonas Gehring, Michael Auli, David Grangier, Denis Yarats, and Yann N. Dauphin. Convolu- tional sequence to sequence learning. arXiv preprint arXiv:1705.03122v2 , 2017." }, { "bbox": { "b": 543.4260864257812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 125.94074249267578, "t": 552.3326416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "[10]" }, { "bbox": { "b": 532.51708984375, "coord_origin": "BOTTOMLEFT", "l": 128.6329345703125, "r": 503.9955139160156, "t": 552.3326416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Alex Graves. Generating sequences with recurrent neural networks. arXiv preprint arXiv:1308.0850 , 2013." }, { "bbox": { "b": 512.716064453125, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.46177673339844, "t": 521.6226196289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "[11]" }, { "bbox": { "b": 490.8970947265625, "coord_origin": "BOTTOMLEFT", "l": 128.0821075439453, "r": 505.6487731933594, "t": 521.6226196289062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for im- age recognition. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 770-778, 2016." }, { "bbox": { "b": 471.0960998535156, "coord_origin": "BOTTOMLEFT", "l": 108.00000762939453, "r": 124.84686279296875, "t": 480.0026550292969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "[12]" }, { "bbox": { "b": 460.18707275390625, "coord_origin": "BOTTOMLEFT", "l": 127.3748779296875, "r": 504.001953125, "t": 480.0026550292969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Sepp Hochreiter, Yoshua Bengio, Paolo Frasconi, and Jürgen Schmidhuber. Gradient flow in recurrent nets: the difficulty of learning long-term dependencies, 2001." }, { "bbox": { "b": 440.3860778808594, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.3443603515625, "t": 449.2926330566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "[13]" }, { "bbox": { "b": 429.4770812988281, "coord_origin": "BOTTOMLEFT", "l": 127.94705963134766, "r": 505.2454528808594, "t": 449.2926330566406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory. Neural computation , 9(8):1735-1780, 1997." }, { "bbox": { "b": 409.67608642578125, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.0400619506836, "t": 418.5826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "[14]" }, { "bbox": { "b": 387.85809326171875, "coord_origin": "BOTTOMLEFT", "l": 127.59709930419922, "r": 504.0025939941406, "t": 418.5826416015625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "Zhongqiang Huang and Mary Harper. Self-training PCFG grammars with latent annotations across languages. In Proceedings of the 2009 Conference on Empirical Methods in Natural Language Processing , pages 832-841. ACL, August 2009." }, { "bbox": { "b": 368.0570983886719, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.76412963867188, "t": 376.9636535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "[15]" }, { "bbox": { "b": 357.1480712890625, "coord_origin": "BOTTOMLEFT", "l": 127.27974700927734, "r": 503.999755859375, "t": 376.9636535644531 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "Rafal Jozefowicz, Oriol Vinyals, Mike Schuster, Noam Shazeer, and Yonghui Wu. Exploring the limits of language modeling. arXiv preprint arXiv:1602.02410 , 2016." }, { "bbox": { "b": 337.3470764160156, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.39153289794922, "t": 346.2536315917969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "[16]" }, { "bbox": { "b": 326.4380798339844, "coord_origin": "BOTTOMLEFT", "l": 126.85124206542969, "r": 503.9952392578125, "t": 346.2536315917969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "Łukasz Kaiser and Samy Bengio. Can active memory replace attention? In Advances in Neural Information Processing Systems, (NIPS) , 2016." }, { "bbox": { "b": 306.6370849609375, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 124.46515655517578, "t": 315.54364013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "[17]" }, { "bbox": { "b": 295.72808837890625, "coord_origin": "BOTTOMLEFT", "l": 126.93594360351562, "r": 504.0040283203125, "t": 315.54364013671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "Łukasz Kaiser and Ilya Sutskever. Neural GPUs learn algorithms. In International Conference on Learning Representations (ICLR) , 2016." }, { "bbox": { "b": 275.9270935058594, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.2394790649414, "t": 284.8336486816406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "[18]" }, { "bbox": { "b": 254.10906982421875, "coord_origin": "BOTTOMLEFT", "l": 126.67637634277344, "r": 505.6534729003906, "t": 284.8336486816406 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "Nal Kalchbrenner, Lasse Espeholt, Karen Simonyan, Aaron van den Oord, Alex Graves, and Ko- ray Kavukcuoglu. Neural machine translation in linear time. arXiv preprint arXiv:1610.10099v2 , 2017." }, { "bbox": { "b": 234.30807495117188, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 124.46531677246094, "t": 243.21463012695312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "[19]" }, { "bbox": { "b": 223.39907836914062, "coord_origin": "BOTTOMLEFT", "l": 126.93611145019531, "r": 505.747314453125, "t": 243.21463012695312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "Yoon Kim, Carl Denton, Luong Hoang, and Alexander M. Rush. Structured attention networks. In International Conference on Learning Representations , 2017." }, { "bbox": { "b": 203.59808349609375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.6935806274414, "t": 212.504638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "[20]" }, { "bbox": { "b": 203.59707641601562, "coord_origin": "BOTTOMLEFT", "l": 127.19862365722656, "r": 505.7453308105469, "t": 212.504638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "Diederik Kingma and Jimmy Ba. Adam: A method for stochastic optimization. In ICLR , 2015." }, { "bbox": { "b": 183.79608154296875, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.598876953125, "t": 192.70263671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 16, "text": "[21]" }, { "bbox": { "b": 172.8870849609375, "coord_origin": "BOTTOMLEFT", "l": 127.08970642089844, "r": 504.0007019042969, "t": 192.70263671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 17, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 16, "text": "Oleksii Kuchaiev and Boris Ginsburg. Factorization tricks for LSTM networks. arXiv preprint arXiv:1703.10722 , 2017." }, { "bbox": { "b": 153.08609008789062, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 126.07655334472656, "t": 161.99264526367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 17, "text": "[22]" }, { "bbox": { "b": 131.26808166503906, "coord_origin": "BOTTOMLEFT", "l": 128.7891082763672, "r": 504.0025939941406, "t": 161.99264526367188 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 18, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 17, "text": "Zhouhan Lin, Minwei Feng, Cicero Nogueira dos Santos, Mo Yu, Bing Xiang, Bowen Zhou, and Yoshua Bengio. A structured self-attentive sentence embedding. arXiv preprint arXiv:1703.03130 , 2017." }, { "bbox": { "b": 111.46707153320312, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 124.56388092041016, "t": 120.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 18, "text": "[23]" }, { "bbox": { "b": 100.55806732177734, "coord_origin": "BOTTOMLEFT", "l": 127.04947662353516, "r": 504.25091552734375, "t": 120.37364196777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 19, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 18, "text": "Minh-Thang Luong, Quoc V. Le, Ilya Sutskever, Oriol Vinyals, and Lukasz Kaiser. Multi-task sequence to sequence learning. arXiv preprint arXiv:1511.06114 , 2015." }, { "bbox": { "b": 80.75707244873047, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.32675170898438, "t": 89.66363525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 19, "text": "[24]" }, { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 126.77674102783203, "r": 505.6534118652344, "t": 89.66363525390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 20, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 19, "text": "Minh-Thang Luong, Hieu Pham, and Christopher D Manning. Effective approaches to attention- arXiv preprint arXiv:1508.04025 , 2015." }, { "bbox": null, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 20, "text": "" }, { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 129.57899475097656, "r": 262.9383544921875, "t": 78.754638671875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 21, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 20, "text": "based neural machine translation." } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 69.84807586669922, "coord_origin": "BOTTOMLEFT", "l": 107.47520446777344, "r": 505.8876953125, "t": 718.3776245117188 }, "charspan": [ 0, 0 ], "page_no": 11 } ], "references": [], "self_ref": "#/tables/4" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 707.8850708007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.181640625, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "[25]" }, { "bbox": { "b": 696.97607421875, "coord_origin": "BOTTOMLEFT", "l": 126.60986328125, "r": 504.00341796875, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Mitchell P Marcus, Mary Ann Marcinkiewicz, and Beatrice Santorini. Building a large annotated corpus of english: The penn treebank. Computational linguistics , 19(2):313-330, 1993." } ], [ { "bbox": { "b": 673.4931030273438, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.77597045898438, "t": 682.399658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "[26]" }, { "bbox": { "b": 651.6751098632812, "coord_origin": "BOTTOMLEFT", "l": 127.29337310791016, "r": 505.2452087402344, "t": 682.399658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "David McClosky, Eugene Charniak, and Mark Johnson. Effective self-training for parsing. In Proceedings of the Human Language Technology Conference of the NAACL, Main Conference , pages 152-159. ACL, June 2006." } ], [ { "bbox": { "b": 628.192138671875, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 124.38294982910156, "t": 637.0986938476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "[27]" }, { "bbox": { "b": 617.2830810546875, "coord_origin": "BOTTOMLEFT", "l": 126.84137725830078, "r": 504.0034484863281, "t": 637.0986938476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Ankur Parikh, Oscar Täckström, Dipanjan Das, and Jakob Uszkoreit. A decomposable attention model. In Empirical Methods in Natural Language Processing , 2016." } ], [ { "bbox": { "b": 593.8001098632812, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.68134307861328, "t": 602.7066650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "[28]" }, { "bbox": { "b": 582.89111328125, "coord_origin": "BOTTOMLEFT", "l": 127.1845474243164, "r": 503.99725341796875, "t": 602.7066650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "Romain Paulus, Caiming Xiong, and Richard Socher. A deep reinforced model for abstractive summarization. arXiv preprint arXiv:1705.04304 , 2017." } ], [ { "bbox": { "b": 559.4081420898438, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 125.52971649169922, "t": 568.314697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "[29]" }, { "bbox": { "b": 526.6810913085938, "coord_origin": "BOTTOMLEFT", "l": 128.16024780273438, "r": 505.2423095703125, "t": 568.314697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "Slav Petrov, Leon Barrett, Romain Thibaux, and Dan Klein. Learning accurate, compact, and interpretable tree annotation. In Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL , pages 433-440. ACL, July 2006." } ], [ { "bbox": { "b": 503.1980895996094, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.23204040527344, "t": 512.1046752929688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "[30]" }, { "bbox": { "b": 492.2890930175781, "coord_origin": "BOTTOMLEFT", "l": 127.81788635253906, "r": 504.0006103515625, "t": 512.1046752929688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Ofir Press and Lior Wolf. Using the output embedding to improve language models. arXiv preprint arXiv:1608.05859 , 2016." } ], [ { "bbox": { "b": 468.80609130859375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.5897216796875, "t": 477.712646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "[31]" }, { "bbox": { "b": 457.8970947265625, "coord_origin": "BOTTOMLEFT", "l": 127.07917022705078, "r": 504.0023498535156, "t": 477.712646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "Rico Sennrich, Barry Haddow, and Alexandra Birch. Neural machine translation of rare words with subword units. arXiv preprint arXiv:1508.07909 , 2015." } ], [ { "bbox": { "b": 434.4140930175781, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.48110961914062, "t": 443.3206481933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "[32]" }, { "bbox": { "b": 412.5960693359375, "coord_origin": "BOTTOMLEFT", "l": 126.95426940917969, "r": 505.248779296875, "t": 443.3206481933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer. arXiv preprint arXiv:1701.06538 , 2017." } ], [ { "bbox": { "b": 389.1130676269531, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.49577331542969, "t": 398.0196228027344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "[33]" }, { "bbox": { "b": 367.2950744628906, "coord_origin": "BOTTOMLEFT", "l": 126.97113037109375, "r": 505.651123046875, "t": 398.0196228027344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Nitish Srivastava, Geoffrey E Hinton, Alex Krizhevsky, Ilya Sutskever, and Ruslan Salakhutdi- nov. Dropout: a simple way to prevent neural networks from overfitting. Journal of Machine Learning Research , 15(1):1929-1958, 2014." } ], [ { "bbox": { "b": 343.81207275390625, "coord_origin": "BOTTOMLEFT", "l": 108.00000762939453, "r": 125.3666763305664, "t": 352.7186279296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "[34]" }, { "bbox": { "b": 311.0840759277344, "coord_origin": "BOTTOMLEFT", "l": 127.97271728515625, "r": 505.245849609375, "t": 352.7186279296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "Sainbayar Sukhbaatar, Arthur Szlam, Jason Weston, and Rob Fergus. End-to-end memory networks. In C. Cortes, N. D. Lawrence, D. D. Lee, M. Sugiyama, and R. Garnett, editors, Advances in Neural Information Processing Systems 28 , pages 2440-2448. Curran Associates, Inc., 2015." } ], [ { "bbox": { "b": 287.60107421875, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.10279083251953, "t": 296.50762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "[35]" }, { "bbox": { "b": 276.69207763671875, "coord_origin": "BOTTOMLEFT", "l": 127.66924285888672, "r": 504.00262451171875, "t": 296.50762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "Ilya Sutskever, Oriol Vinyals, and Quoc VV Le. Sequence to sequence learning with neural networks. In Advances in Neural Information Processing Systems , pages 3104-3112, 2014." } ], [ { "bbox": { "b": 253.20907592773438, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.82581329345703, "t": 262.1156311035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "[36]" }, { "bbox": { "b": 242.30007934570312, "coord_origin": "BOTTOMLEFT", "l": 127.35070037841797, "r": 505.74029541015625, "t": 262.1156311035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens, and Zbigniew Wojna. Rethinking the inception architecture for computer vision. CoRR , abs/1512.00567, 2015." } ], [ { "bbox": { "b": 218.81707763671875, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 125.31416320800781, "t": 227.7236328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "[37]" }, { "bbox": { "b": 207.9080810546875, "coord_origin": "BOTTOMLEFT", "l": 127.9123306274414, "r": 504.00244140625, "t": 227.7236328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "Vinyals & Kaiser, Koo, Petrov, Sutskever, and Hinton. Grammar as a foreign language. In Advances in Neural Information Processing Systems , 2015." } ], [ { "bbox": { "b": 184.42507934570312, "coord_origin": "BOTTOMLEFT", "l": 108.0000228881836, "r": 125.4793472290039, "t": 193.33163452148438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "[38]" }, { "bbox": { "b": 151.69808959960938, "coord_origin": "BOTTOMLEFT", "l": 128.10232543945312, "r": 504.0032653808594, "t": 193.33163452148438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun, Yuan Cao, Qin Gao, Klaus Macherey, et al. Google's neural machine translation system: Bridging the gap between human and machine translation. arXiv preprint arXiv:1609.08144 , 2016." } ], [ { "bbox": { "b": 128.215087890625, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 125.64002990722656, "t": 137.12164306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "[39]" }, { "bbox": { "b": 117.30608367919922, "coord_origin": "BOTTOMLEFT", "l": 128.287109375, "r": 504.00250244140625, "t": 137.12164306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "Jie Zhou, Ying Cao, Xuguang Wang, Peng Li, and Wei Xu. Deep recurrent models with fast-forward connections for neural machine translation. CoRR , abs/1606.04199, 2016." } ], [ { "bbox": { "b": 93.82307434082031, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 125.6090087890625, "t": 102.7296371459961 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "[40]" }, { "bbox": { "b": 72.00507354736328, "coord_origin": "BOTTOMLEFT", "l": 128.2514190673828, "r": 504.0024719238281, "t": 102.7296371459961 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "Muhua Zhu, Yue Zhang, Wenliang Chen, Min Zhang, and Jingbo Zhu. Fast and accurate shift-reduce constituent parsing. In Proceedings of the 51st Annual Meeting of the ACL (Volume 1: Long Papers) , pages 434-443. ACL, August 2013." } ] ], "num_cols": 2, "num_rows": 16, "table_cells": [ { "bbox": { "b": 707.8850708007812, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.181640625, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "[25]" }, { "bbox": { "b": 696.97607421875, "coord_origin": "BOTTOMLEFT", "l": 126.60986328125, "r": 504.00341796875, "t": 716.7916259765625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Mitchell P Marcus, Mary Ann Marcinkiewicz, and Beatrice Santorini. Building a large annotated corpus of english: The penn treebank. Computational linguistics , 19(2):313-330, 1993." }, { "bbox": { "b": 673.4931030273438, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.77597045898438, "t": 682.399658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "[26]" }, { "bbox": { "b": 651.6751098632812, "coord_origin": "BOTTOMLEFT", "l": 127.29337310791016, "r": 505.2452087402344, "t": 682.399658203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "David McClosky, Eugene Charniak, and Mark Johnson. Effective self-training for parsing. In Proceedings of the Human Language Technology Conference of the NAACL, Main Conference , pages 152-159. ACL, June 2006." }, { "bbox": { "b": 628.192138671875, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 124.38294982910156, "t": 637.0986938476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "[27]" }, { "bbox": { "b": 617.2830810546875, "coord_origin": "BOTTOMLEFT", "l": 126.84137725830078, "r": 504.0034484863281, "t": 637.0986938476562 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "Ankur Parikh, Oscar Täckström, Dipanjan Das, and Jakob Uszkoreit. A decomposable attention model. In Empirical Methods in Natural Language Processing , 2016." }, { "bbox": { "b": 593.8001098632812, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.68134307861328, "t": 602.7066650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "[28]" }, { "bbox": { "b": 582.89111328125, "coord_origin": "BOTTOMLEFT", "l": 127.1845474243164, "r": 503.99725341796875, "t": 602.7066650390625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "Romain Paulus, Caiming Xiong, and Richard Socher. A deep reinforced model for abstractive summarization. arXiv preprint arXiv:1705.04304 , 2017." }, { "bbox": { "b": 559.4081420898438, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 125.52971649169922, "t": 568.314697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "[29]" }, { "bbox": { "b": 526.6810913085938, "coord_origin": "BOTTOMLEFT", "l": 128.16024780273438, "r": 505.2423095703125, "t": 568.314697265625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "Slav Petrov, Leon Barrett, Romain Thibaux, and Dan Klein. Learning accurate, compact, and interpretable tree annotation. In Proceedings of the 21st International Conference on Computational Linguistics and 44th Annual Meeting of the ACL , pages 433-440. ACL, July 2006." }, { "bbox": { "b": 503.1980895996094, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.23204040527344, "t": 512.1046752929688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "[30]" }, { "bbox": { "b": 492.2890930175781, "coord_origin": "BOTTOMLEFT", "l": 127.81788635253906, "r": 504.0006103515625, "t": 512.1046752929688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "Ofir Press and Lior Wolf. Using the output embedding to improve language models. arXiv preprint arXiv:1608.05859 , 2016." }, { "bbox": { "b": 468.80609130859375, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.5897216796875, "t": 477.712646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "[31]" }, { "bbox": { "b": 457.8970947265625, "coord_origin": "BOTTOMLEFT", "l": 127.07917022705078, "r": 504.0023498535156, "t": 477.712646484375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "Rico Sennrich, Barry Haddow, and Alexandra Birch. Neural machine translation of rare words with subword units. arXiv preprint arXiv:1508.07909 , 2015." }, { "bbox": { "b": 434.4140930175781, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.48110961914062, "t": 443.3206481933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "[32]" }, { "bbox": { "b": 412.5960693359375, "coord_origin": "BOTTOMLEFT", "l": 126.95426940917969, "r": 505.248779296875, "t": 443.3206481933594 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer. arXiv preprint arXiv:1701.06538 , 2017." }, { "bbox": { "b": 389.1130676269531, "coord_origin": "BOTTOMLEFT", "l": 108, "r": 124.49577331542969, "t": 398.0196228027344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "[33]" }, { "bbox": { "b": 367.2950744628906, "coord_origin": "BOTTOMLEFT", "l": 126.97113037109375, "r": 505.651123046875, "t": 398.0196228027344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Nitish Srivastava, Geoffrey E Hinton, Alex Krizhevsky, Ilya Sutskever, and Ruslan Salakhutdi- nov. Dropout: a simple way to prevent neural networks from overfitting. Journal of Machine Learning Research , 15(1):1929-1958, 2014." }, { "bbox": { "b": 343.81207275390625, "coord_origin": "BOTTOMLEFT", "l": 108.00000762939453, "r": 125.3666763305664, "t": 352.7186279296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "[34]" }, { "bbox": { "b": 311.0840759277344, "coord_origin": "BOTTOMLEFT", "l": 127.97271728515625, "r": 505.245849609375, "t": 352.7186279296875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "Sainbayar Sukhbaatar, Arthur Szlam, Jason Weston, and Rob Fergus. End-to-end memory networks. In C. Cortes, N. D. Lawrence, D. D. Lee, M. Sugiyama, and R. Garnett, editors, Advances in Neural Information Processing Systems 28 , pages 2440-2448. Curran Associates, Inc., 2015." }, { "bbox": { "b": 287.60107421875, "coord_origin": "BOTTOMLEFT", "l": 107.99999237060547, "r": 125.10279083251953, "t": 296.50762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "[35]" }, { "bbox": { "b": 276.69207763671875, "coord_origin": "BOTTOMLEFT", "l": 127.66924285888672, "r": 504.00262451171875, "t": 296.50762939453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 10, "text": "Ilya Sutskever, Oriol Vinyals, and Quoc VV Le. Sequence to sequence learning with neural networks. In Advances in Neural Information Processing Systems , pages 3104-3112, 2014." }, { "bbox": { "b": 253.20907592773438, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 124.82581329345703, "t": 262.1156311035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "[36]" }, { "bbox": { "b": 242.30007934570312, "coord_origin": "BOTTOMLEFT", "l": 127.35070037841797, "r": 505.74029541015625, "t": 262.1156311035156 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens, and Zbigniew Wojna. Rethinking the inception architecture for computer vision. CoRR , abs/1512.00567, 2015." }, { "bbox": { "b": 218.81707763671875, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 125.31416320800781, "t": 227.7236328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 12, "text": "[37]" }, { "bbox": { "b": 207.9080810546875, "coord_origin": "BOTTOMLEFT", "l": 127.9123306274414, "r": 504.00244140625, "t": 227.7236328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 13, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 12, "text": "Vinyals & Kaiser, Koo, Petrov, Sutskever, and Hinton. Grammar as a foreign language. In Advances in Neural Information Processing Systems , 2015." }, { "bbox": { "b": 184.42507934570312, "coord_origin": "BOTTOMLEFT", "l": 108.0000228881836, "r": 125.4793472290039, "t": 193.33163452148438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 13, "text": "[38]" }, { "bbox": { "b": 151.69808959960938, "coord_origin": "BOTTOMLEFT", "l": 128.10232543945312, "r": 504.0032653808594, "t": 193.33163452148438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 14, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 13, "text": "Yonghui Wu, Mike Schuster, Zhifeng Chen, Quoc V Le, Mohammad Norouzi, Wolfgang Macherey, Maxim Krikun, Yuan Cao, Qin Gao, Klaus Macherey, et al. Google's neural machine translation system: Bridging the gap between human and machine translation. arXiv preprint arXiv:1609.08144 , 2016." }, { "bbox": { "b": 128.215087890625, "coord_origin": "BOTTOMLEFT", "l": 107.9999771118164, "r": 125.64002990722656, "t": 137.12164306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 14, "text": "[39]" }, { "bbox": { "b": 117.30608367919922, "coord_origin": "BOTTOMLEFT", "l": 128.287109375, "r": 504.00250244140625, "t": 137.12164306640625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 15, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 14, "text": "Jie Zhou, Ying Cao, Xuguang Wang, Peng Li, and Wei Xu. Deep recurrent models with fast-forward connections for neural machine translation. CoRR , abs/1606.04199, 2016." }, { "bbox": { "b": 93.82307434082031, "coord_origin": "BOTTOMLEFT", "l": 107.99996948242188, "r": 125.6090087890625, "t": 102.7296371459961 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 15, "text": "[40]" }, { "bbox": { "b": 72.00507354736328, "coord_origin": "BOTTOMLEFT", "l": 128.2514190673828, "r": 504.0024719238281, "t": 102.7296371459961 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 16, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 15, "text": "Muhua Zhu, Yue Zhang, Wenliang Chen, Min Zhang, and Jingbo Zhu. Fast and accurate shift-reduce constituent parsing. In Proceedings of the 51st Annual Meeting of the ACL (Volume 1: Long Papers) , pages 434-443. ACL, August 2013." } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 72.00507354736328, "coord_origin": "BOTTOMLEFT", "l": 106.71272277832031, "r": 506.05230712890625, "t": 719.6456298828125 }, "charspan": [ 0, 0 ], "page_no": 12 } ], "references": [], "self_ref": "#/tables/5" } ]

[]

{ "1": { "page_no": 1, "size": { "height": 792, "width": 612 } }, "2": { "page_no": 2, "size": { "height": 792, "width": 612 } }, "3": { "page_no": 3, "size": { "height": 792, "width": 612 } }, "4": { "page_no": 4, "size": { "height": 792, "width": 612 } }, "5": { "page_no": 5, "size": { "height": 792, "width": 612 } }, "6": { "page_no": 6, "size": { "height": 792, "width": 612 } }, "7": { "page_no": 7, "size": { "height": 792, "width": 612 } }, "8": { "page_no": 8, "size": { "height": 792, "width": 612 } }, "9": { "page_no": 9, "size": { "height": 792, "width": 612 } }, "10": { "page_no": 10, "size": { "height": 792, "width": 612 } }, "11": { "page_no": 11, "size": { "height": 792, "width": 612 } }, "12": { "page_no": 12, "size": { "height": 792, "width": 612 } }, "13": { "page_no": 13, "size": { "height": 792, "width": 612 } }, "14": { "page_no": 14, "size": { "height": 792, "width": 612 } }, "15": { "page_no": 15, "size": { "height": 792, "width": 612 } }, "16": null, "17": null, "18": null, "19": null, "20": null, "21": null, "22": null, "23": null, "24": null, "25": null, "26": null, "27": null, "28": null, "29": null, "30": null, "31": null, "32": null, "33": null, "34": null, "35": null, "36": null }

{ "children": [], "label": "unspecified", "name": "_root_", "self_ref": "#/furniture" }

{ "children": [ { "$ref": "#/texts/0" }, { "$ref": "#/pictures/0" }, { "$ref": "#/texts/1" }, { "$ref": "#/texts/2" }, { "$ref": "#/texts/3" }, { "$ref": "#/texts/4" }, { "$ref": "#/texts/5" }, { "$ref": "#/texts/6" }, { "$ref": "#/texts/7" }, { "$ref": "#/texts/8" }, { "$ref": "#/texts/9" }, { "$ref": "#/texts/10" }, { "$ref": "#/texts/11" }, { "$ref": "#/texts/12" }, { "$ref": "#/texts/13" }, { "$ref": "#/texts/14" }, { "$ref": "#/texts/15" }, { "$ref": "#/pictures/1" }, { "$ref": "#/texts/16" }, { "$ref": "#/texts/17" }, { "$ref": "#/texts/18" }, { "$ref": "#/texts/19" }, { "$ref": "#/texts/20" }, { "$ref": "#/texts/21" }, { "$ref": "#/texts/22" }, { "$ref": "#/texts/23" }, { "$ref": "#/texts/24" }, { "$ref": "#/tables/0" }, { "$ref": "#/tables/1" }, { "$ref": "#/texts/25" }, { "$ref": "#/pictures/2" }, { "$ref": "#/tables/2" }, { "$ref": "#/texts/26" }, { "$ref": "#/texts/27" }, { "$ref": "#/texts/28" }, { "$ref": "#/texts/29" }, { "$ref": "#/texts/30" }, { "$ref": "#/texts/31" }, { "$ref": "#/texts/32" }, { "$ref": "#/texts/33" }, { "$ref": "#/texts/34" }, { "$ref": "#/texts/35" }, { "$ref": "#/texts/36" }, { "$ref": "#/pictures/3" }, { "$ref": "#/texts/37" }, { "$ref": "#/texts/38" }, { "$ref": "#/texts/39" }, { "$ref": "#/texts/40" }, { "$ref": "#/texts/41" }, { "$ref": "#/pictures/4" }, { "$ref": "#/texts/42" }, { "$ref": "#/texts/43" }, { "$ref": "#/texts/44" }, { "$ref": "#/texts/45" }, { "$ref": "#/texts/46" }, { "$ref": "#/texts/47" }, { "$ref": "#/texts/48" }, { "$ref": "#/texts/49" }, { "$ref": "#/texts/50" }, { "$ref": "#/texts/51" }, { "$ref": "#/texts/52" }, { "$ref": "#/texts/53" }, { "$ref": "#/texts/54" }, { "$ref": "#/texts/55" }, { "$ref": "#/texts/56" }, { "$ref": "#/texts/57" }, { "$ref": "#/texts/58" }, { "$ref": "#/pictures/5" }, { "$ref": "#/texts/59" }, { "$ref": "#/texts/60" }, { "$ref": "#/texts/61" }, { "$ref": "#/texts/62" }, { "$ref": "#/texts/63" }, { "$ref": "#/texts/64" }, { "$ref": "#/texts/65" }, { "$ref": "#/groups/0" }, { "$ref": "#/texts/68" }, { "$ref": "#/texts/69" }, { "$ref": "#/texts/70" }, { "$ref": "#/texts/71" }, { "$ref": "#/texts/72" }, { "$ref": "#/texts/73" }, { "$ref": "#/texts/74" }, { "$ref": "#/texts/75" }, { "$ref": "#/texts/76" }, { "$ref": "#/texts/77" }, { "$ref": "#/texts/78" }, { "$ref": "#/texts/79" }, { "$ref": "#/texts/80" }, { "$ref": "#/texts/81" }, { "$ref": "#/texts/82" }, { "$ref": "#/texts/83" }, { "$ref": "#/texts/84" }, { "$ref": "#/texts/85" }, { "$ref": "#/texts/86" }, { "$ref": "#/texts/87" }, { "$ref": "#/texts/88" }, { "$ref": "#/texts/89" }, { "$ref": "#/texts/90" }, { "$ref": "#/texts/91" }, { "$ref": "#/texts/92" }, { "$ref": "#/texts/93" }, { "$ref": "#/texts/94" }, { "$ref": "#/texts/95" }, { "$ref": "#/texts/96" }, { "$ref": "#/texts/97" }, { "$ref": "#/texts/98" }, { "$ref": "#/texts/99" }, { "$ref": "#/texts/100" }, { "$ref": "#/texts/101" }, { "$ref": "#/texts/102" }, { "$ref": "#/texts/103" }, { "$ref": "#/texts/104" }, { "$ref": "#/texts/105" }, { "$ref": "#/texts/106" }, { "$ref": "#/texts/107" }, { "$ref": "#/texts/108" }, { "$ref": "#/texts/109" }, { "$ref": "#/texts/110" }, { "$ref": "#/texts/111" }, { "$ref": "#/texts/112" } ], "label": "unspecified", "name": "_root_", "self_ref": "#/body" }

[ { "children": [ { "$ref": "#/texts/66" }, { "$ref": "#/texts/67" } ], "label": "list", "name": "list", "parent": { "$ref": "#/body" }, "self_ref": "#/groups/0" } ]

[ { "children": [], "enumerated": null, "label": "page_header", "level": null, "marker": null, "orig": "arXiv:1910.03771v5 [cs.CL] 14 Jul 2020", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 231.99996948242188, "coord_origin": "BOTTOMLEFT", "l": 17.635372161865234, "r": 36.339786529541016, "t": 572.5114135742188 }, "charspan": [ 0, 38 ], "page_no": 1 } ], "self_ref": "#/texts/0", "text": "arXiv:1910.03771v5 [cs.CL] 14 Jul 2020" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Transformers: State-of-the-Art Natural Language Processing", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 753.6815795898438, "coord_origin": "BOTTOMLEFT", "l": 125.16598510742188, "r": 501.1102294921875, "t": 768.1585693359375 }, "charspan": [ 0, 58 ], "page_no": 1 } ], "self_ref": "#/texts/1", "text": "Transformers: State-of-the-Art Natural Language Processing" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, R'emi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lhoest, Alexander M. Rush", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 677.3466186523438, "coord_origin": "BOTTOMLEFT", "l": 91.74420166015625, "r": 507.9783020019531, "t": 731.5494995117188 }, "charspan": [ 0, 322 ], "page_no": 1 } ], "self_ref": "#/texts/2", "text": "Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, R'emi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lhoest, Alexander M. Rush" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hugging Face, Brooklyn, USA / { first-name } @huggingface.co", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 649.442138671875, "coord_origin": "BOTTOMLEFT", "l": 125.29016876220703, "r": 474.6858215332031, "t": 661.850341796875 }, "charspan": [ 0, 60 ], "page_no": 1 } ], "self_ref": "#/texts/3", "text": "Hugging Face, Brooklyn, USA / { first-name } @huggingface.co" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "Abstract", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 605.8912963867188, "coord_origin": "BOTTOMLEFT", "l": 158.354248046875, "r": 203.37631225585938, "t": 617.5703125 }, "charspan": [ 0, 8 ], "page_no": 1 } ], "self_ref": "#/texts/4", "text": "Abstract" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Recent progress in natural language processing has been driven by advances in both model architecture and model pretraining. Transformer architectures have facilitated building higher-capacity models and pretraining has made it possible to effectively utilize this capacity for a wide variety of tasks. Transformers is an open-source library with the goal of opening up these advances to the wider machine learning community. The library consists of carefully engineered stateof-the art Transformer architectures under a unified API. Backing this library is a curated collection of pretrained models made by and available for the community. Transformers is designed to be extensible by researchers, simple for practitioners, and fast and robust in industrial deployments. The library is available at https://github.com/ huggingface/transformers .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 355.2526550292969, "coord_origin": "BOTTOMLEFT", "l": 87.46231079101562, "r": 274.91387939453125, "t": 592.04833984375 }, "charspan": [ 0, 846 ], "page_no": 1 } ], "self_ref": "#/texts/5", "text": "Recent progress in natural language processing has been driven by advances in both model architecture and model pretraining. Transformer architectures have facilitated building higher-capacity models and pretraining has made it possible to effectively utilize this capacity for a wide variety of tasks. Transformers is an open-source library with the goal of opening up these advances to the wider machine learning community. The library consists of carefully engineered stateof-the art Transformer architectures under a unified API. Backing this library is a curated collection of pretrained models made by and available for the community. Transformers is designed to be extensible by researchers, simple for practitioners, and fast and robust in industrial deployments. The library is available at https://github.com/ huggingface/transformers ." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "1 Introduction", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 329.1354064941406, "coord_origin": "BOTTOMLEFT", "l": 71.94170379638672, "r": 154.81369018554688, "t": 340.9704895019531 }, "charspan": [ 0, 14 ], "page_no": 1 } ], "self_ref": "#/texts/6", "text": "1 Introduction" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The Transformer (Vaswani et al., 2017) has rapidly become the dominant architecture for natural language processing, surpassing alternative neural models such as convolutional and recurrent neural networks in performance for tasks in both natural language understanding and natural language generation. The architecture scales with training data and model size, facilitates efficient parallel training, and captures long-range sequence features.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 198.25860595703125, "coord_origin": "BOTTOMLEFT", "l": 71.0051040649414, "r": 292.08306884765625, "t": 318.21197509765625 }, "charspan": [ 0, 445 ], "page_no": 1 } ], "self_ref": "#/texts/7", "text": "The Transformer (Vaswani et al., 2017) has rapidly become the dominant architecture for natural language processing, surpassing alternative neural models such as convolutional and recurrent neural networks in performance for tasks in both natural language understanding and natural language generation. The architecture scales with training data and model size, facilitates efficient parallel training, and captures long-range sequence features." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Model pretraining (McCann et al., 2017; Howard and Ruder, 2018; Peters et al., 2018; Devlin et al., 2018) allows models to be trained on generic corpora and subsequently be easily adapted to specific tasks with strong performance. The Transformer architecture is particularly conducive to pretraining on large text corpora, leading to major gains in accuracy on downstream tasks including text classification (Yang et al., 2019), language understanding", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.89776611328125, "coord_origin": "BOTTOMLEFT", "l": 71.32652282714844, "r": 292.0832214355469, "t": 195.742431640625 }, "charspan": [ 0, 452 ], "page_no": 1 } ], "self_ref": "#/texts/8", "text": "Model pretraining (McCann et al., 2017; Howard and Ruder, 2018; Peters et al., 2018; Devlin et al., 2018) allows models to be trained on generic corpora and subsequently be easily adapted to specific tasks with strong performance. The Transformer architecture is particularly conducive to pretraining on large text corpora, leading to major gains in accuracy on downstream tasks including text classification (Yang et al., 2019), language understanding" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "(Liu et al., 2019b; Wang et al., 2018, 2019), machine translation (Lample and Conneau, 2019a), coreference resolution (Joshi et al., 2019), commonsense inference (Bosselut et al., 2019), and summarization (Lewis et al., 2019) among others.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 551.5374755859375, "coord_origin": "BOTTOMLEFT", "l": 306.3004150390625, "r": 527.3585815429688, "t": 617.0319213867188 }, "charspan": [ 0, 239 ], "page_no": 1 } ], "self_ref": "#/texts/9", "text": "(Liu et al., 2019b; Wang et al., 2018, 2019), machine translation (Lample and Conneau, 2019a), coreference resolution (Joshi et al., 2019), commonsense inference (Bosselut et al., 2019), and summarization (Lewis et al., 2019) among others." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "This advance leads to a wide range of practical challenges that must be addressed in order for these models to be widely utilized. The ubiquitous use of the Transformer calls for systems to train, analyze, scale, and augment the model on a variety of platforms. The architecture is used as a building block to design increasingly sophisticated extensions and precise experiments. The pervasive adoption of pretraining methods has led to the need to distribute, fine-tune, deploy, and compress the core pretrained models used by the community.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 401.6963195800781, "coord_origin": "BOTTOMLEFT", "l": 306.33514404296875, "r": 527.3592529296875, "t": 548.81201171875 }, "charspan": [ 0, 542 ], "page_no": 1 } ], "self_ref": "#/texts/10", "text": "This advance leads to a wide range of practical challenges that must be addressed in order for these models to be widely utilized. The ubiquitous use of the Transformer calls for systems to train, analyze, scale, and augment the model on a variety of platforms. The architecture is used as a building block to design increasingly sophisticated extensions and precise experiments. The pervasive adoption of pretraining methods has led to the need to distribute, fine-tune, deploy, and compress the core pretrained models used by the community." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Transformers is a library dedicated to supporting Transformer-based architectures and facilitating the distribution of pretrained models. At the core of the libary is an implementation of the Transformer which is designed for both research and production. The philosophy is to support industrial-strength implementations of popular model variants that are easy to read, extend, and deploy. On this foundation, the library supports the distribution and usage of a wide-variety of pretrained models in a centralized model hub. This hub supports users to compare different models with the same minimal API and to experiment with shared models on a variety of different tasks.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 212.63763427734375, "coord_origin": "BOTTOMLEFT", "l": 306.0239562988281, "r": 527.4473266601562, "t": 399.2933044433594 }, "charspan": [ 0, 672 ], "page_no": 1 } ], "self_ref": "#/texts/11", "text": "Transformers is a library dedicated to supporting Transformer-based architectures and facilitating the distribution of pretrained models. At the core of the libary is an implementation of the Transformer which is designed for both research and production. The philosophy is to support industrial-strength implementations of popular model variants that are easy to read, extend, and deploy. On this foundation, the library supports the distribution and usage of a wide-variety of pretrained models in a centralized model hub. This hub supports users to compare different models with the same minimal API and to experiment with shared models on a variety of different tasks." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Transformers is an ongoing effort maintained by the team of engineers and researchers at Hugging Face with support from a vibrant community of over 400 external contributors. The library is released under the Apache 2.0 license and is available on GitHub$^{1}$. Detailed documentation and tutorials are available on Hugging Face's website$^{2}$.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 117.18121337890625, "coord_origin": "BOTTOMLEFT", "l": 306.07269287109375, "r": 527.3513793945312, "t": 209.60992431640625 }, "charspan": [ 0, 345 ], "page_no": 1 } ], "self_ref": "#/texts/12", "text": "Transformers is an ongoing effort maintained by the team of engineers and researchers at Hugging Face with support from a vibrant community of over 400 external contributors. The library is released under the Apache 2.0 license and is available on GitHub$^{1}$. Detailed documentation and tutorials are available on Hugging Face's website$^{2}$." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{1}$https://github.com/huggingface/ transformers", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 88.04669189453125, "coord_origin": "BOTTOMLEFT", "l": 306.96014404296875, "r": 490.1900634765625, "t": 107.353759765625 }, "charspan": [ 0, 50 ], "page_no": 1 } ], "self_ref": "#/texts/13", "text": "$^{1}$https://github.com/huggingface/ transformers" }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{2}$https://huggingface.co/transformers/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 76.14422607421875, "coord_origin": "BOTTOMLEFT", "l": 318.8753356933594, "r": 517.0892333984375, "t": 85.55902099609375 }, "charspan": [ 0, 42 ], "page_no": 1 } ], "self_ref": "#/texts/14", "text": "$^{2}$https://huggingface.co/transformers/" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 1: Average daily unique downloads of the most downloaded pretrained models, Oct. 2019 to May 2020.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 592.0045166015625, "coord_origin": "BOTTOMLEFT", "l": 76.58876037597656, "r": 520.1817016601562, "t": 602.2489624023438 }, "charspan": [ 0, 105 ], "page_no": 2 } ], "self_ref": "#/texts/15", "text": "Figure 1: Average daily unique downloads of the most downloaded pretrained models, Oct. 2019 to May 2020." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "2 Related Work", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 557.4793701171875, "coord_origin": "BOTTOMLEFT", "l": 71.3844985961914, "r": 161.312255859375, "t": 568.9392700195312 }, "charspan": [ 0, 14 ], "page_no": 2 } ], "self_ref": "#/texts/16", "text": "2 Related Work" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The NLP and ML communities have a strong culture of building open-source research tools. The structure of Transformers is inspired by the pioneering tensor2tensor library (Vaswani et al., 2018) and the original source code for BERT (Devlin et al., 2018), both from Google Research. The concept of providing easy caching for pretrained models stemmed from AllenNLP (Gardner et al., 2018). The library is also closely related to neural translation and language modeling systems, such as Fairseq (Ott et al., 2019), OpenNMT (Klein et al., 2017), Texar (Hu et al., 2018), Megatron-LM (Shoeybi et al., 2019), and Marian NMT (Junczys-Dowmunt et al., 2018). Building on these elements, Transformers adds extra user-facing features to allow for easy downloading, caching, and fine-tuning of the models as well as seamless transition to production. Transformers maintains some compatibility with these libraries, most directly including a tool for performing inference using models from Marian NMT and Google's BERT.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 252.70762634277344, "coord_origin": "BOTTOMLEFT", "l": 70.87841033935547, "r": 292.1806640625, "t": 548.2826538085938 }, "charspan": [ 0, 1007 ], "page_no": 2 } ], "self_ref": "#/texts/17", "text": "The NLP and ML communities have a strong culture of building open-source research tools. The structure of Transformers is inspired by the pioneering tensor2tensor library (Vaswani et al., 2018) and the original source code for BERT (Devlin et al., 2018), both from Google Research. The concept of providing easy caching for pretrained models stemmed from AllenNLP (Gardner et al., 2018). The library is also closely related to neural translation and language modeling systems, such as Fairseq (Ott et al., 2019), OpenNMT (Klein et al., 2017), Texar (Hu et al., 2018), Megatron-LM (Shoeybi et al., 2019), and Marian NMT (Junczys-Dowmunt et al., 2018). Building on these elements, Transformers adds extra user-facing features to allow for easy downloading, caching, and fine-tuning of the models as well as seamless transition to production. Transformers maintains some compatibility with these libraries, most directly including a tool for performing inference using models from Marian NMT and Google's BERT." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "There is a long history of easy-to-use, userfacing libraries for general-purpose NLP. Two core libraries are NLTK (Loper and Bird, 2002) and Stanford CoreNLP (Manning et al., 2014), which collect a variety of different approaches to NLP in a single package. More recently, general-purpose, open-source libraries have focused primarily on machine learning for a variety of NLP tasks, these include Spacy (Honnibal and Montani, 2017), AllenNLP (Gardner et al., 2018), flair (Akbik et al., 2019), and Stanza (Qi et al., 2020). Transformers provides similar functionality as these libraries. Additionally, each of these libraries now uses the", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.908447265625, "coord_origin": "BOTTOMLEFT", "l": 71.11236572265625, "r": 292.17230224609375, "t": 249.3594970703125 }, "charspan": [ 0, 638 ], "page_no": 2 } ], "self_ref": "#/texts/18", "text": "There is a long history of easy-to-use, userfacing libraries for general-purpose NLP. Two core libraries are NLTK (Loper and Bird, 2002) and Stanford CoreNLP (Manning et al., 2014), which collect a variety of different approaches to NLP in a single package. More recently, general-purpose, open-source libraries have focused primarily on machine learning for a variety of NLP tasks, these include Spacy (Honnibal and Montani, 2017), AllenNLP (Gardner et al., 2018), flair (Akbik et al., 2019), and Stanza (Qi et al., 2020). Transformers provides similar functionality as these libraries. Additionally, each of these libraries now uses the" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Transformers library and model hub as a low-level framework.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 544.0715942382812, "coord_origin": "BOTTOMLEFT", "l": 306.32318115234375, "r": 525.5380249023438, "t": 568.1483764648438 }, "charspan": [ 0, 60 ], "page_no": 2 } ], "self_ref": "#/texts/19", "text": "Transformers library and model hub as a low-level framework." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Since Transformers provides a hub for NLP models, it is also related to popular model hubs including Torch Hub and TensorFlow Hub which collect framework-specific model parameters for easy use. Unlike these hubs, Transformers is domain-specific which allows the system to provide automatic support for model analysis, usage, deployment, benchmarking, and easy replicability.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 434.085205078125, "coord_origin": "BOTTOMLEFT", "l": 306.4312438964844, "r": 527.44921875, "t": 540.2442626953125 }, "charspan": [ 0, 374 ], "page_no": 2 } ], "self_ref": "#/texts/20", "text": "Since Transformers provides a hub for NLP models, it is also related to popular model hubs including Torch Hub and TensorFlow Hub which collect framework-specific model parameters for easy use. Unlike these hubs, Transformers is domain-specific which allows the system to provide automatic support for model analysis, usage, deployment, benchmarking, and easy replicability." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "3 Library Design", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 407.8183898925781, "coord_origin": "BOTTOMLEFT", "l": 306.27239990234375, "r": 403.2523498535156, "t": 419.6124572753906 }, "charspan": [ 0, 16 ], "page_no": 2 } ], "self_ref": "#/texts/21", "text": "3 Library Design" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Transformers is designed to mirror the standard NLP machine learning model pipeline: process data, apply a model, and make predictions. Although the library includes tools facilitating training and development, in this technical report we focus on the core modeling specifications. For complete details about the features of the library refer to the documentation available on https: //huggingface.co/transformers/ .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 277.1366271972656, "coord_origin": "BOTTOMLEFT", "l": 306.02349853515625, "r": 528.2330322265625, "t": 396.3193054199219 }, "charspan": [ 0, 416 ], "page_no": 2 } ], "self_ref": "#/texts/22", "text": "Transformers is designed to mirror the standard NLP machine learning model pipeline: process data, apply a model, and make predictions. Although the library includes tools facilitating training and development, in this technical report we focus on the core modeling specifications. For complete details about the features of the library refer to the documentation available on https: //huggingface.co/transformers/ ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Every model in the library is fully defined by three building blocks shown in the diagram in Figure 2: (a) a tokenizer, which converts raw text to sparse index encodings, (b) a transformer, which transforms sparse indices to contextual embeddings, and (c) a head, which uses contextual embeddings to make a task-specific prediction. Most user needs can be addressed with these three components.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 154.3603515625, "coord_origin": "BOTTOMLEFT", "l": 306.34820556640625, "r": 527.359375, "t": 273.66119384765625 }, "charspan": [ 0, 394 ], "page_no": 2 } ], "self_ref": "#/texts/23", "text": "Every model in the library is fully defined by three building blocks shown in the diagram in Figure 2: (a) a tokenizer, which converts raw text to sparse index encodings, (b) a transformer, which transforms sparse indices to contextual embeddings, and (c) a head, which uses contextual embeddings to make a task-specific prediction. Most user needs can be addressed with these three components." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Transformers Central to the library are carefully tested implementations of Transformer architecture variants which are widely used in NLP. The full list of currently implemented architectures is shown in Figure 2 (Left). While each of these architectures", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.77947998046875, "coord_origin": "BOTTOMLEFT", "l": 306.2636413574219, "r": 525.9222412109375, "t": 141.69683837890625 }, "charspan": [ 0, 255 ], "page_no": 2 } ], "self_ref": "#/texts/24", "text": "Transformers Central to the library are carefully tested implementations of Transformer architecture variants which are widely used in NLP. The full list of currently implemented architectures is shown in Figure 2 (Left). While each of these architectures" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 2: The Transformers library. (Diagram-Right) Each model is made up of a Tokenizer, Transformer, and Head. The model is pretrained with a fixed head and can then be further fine-tuned with alternate heads for different tasks. (Bottom) Each model uses a specific Tokenizer either implemented in Python or in Rust. These often differ in small details, but need to be in sync with pretraining. (Left) Transformer architectures specialized for different tasks, e.g. understanding versus generation, or for specific use-cases, e.g. speed, image+text. (Top) heads allow a Transformer to be used for different tasks. Here we assume the input token sequence is x$_{1:}$$_{N}$ from a vocabulary V , and y represents different possible outputs, possibly from a class set C . Example datasets represent a small subset of example code distributed with the library.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 67.82977294921875, "coord_origin": "BOTTOMLEFT", "l": 70.88860321044922, "r": 526.7886962890625, "t": 162.25830078125 }, "charspan": [ 0, 858 ], "page_no": 3 } ], "self_ref": "#/texts/25", "text": "Figure 2: The Transformers library. (Diagram-Right) Each model is made up of a Tokenizer, Transformer, and Head. The model is pretrained with a fixed head and can then be further fine-tuned with alternate heads for different tasks. (Bottom) Each model uses a specific Tokenizer either implemented in Python or in Rust. These often differ in small details, but need to be in sync with pretraining. (Left) Transformer architectures specialized for different tasks, e.g. understanding versus generation, or for specific use-cases, e.g. speed, image+text. (Top) heads allow a Transformer to be used for different tasks. Here we assume the input token sequence is x$_{1:}$$_{N}$ from a vocabulary V , and y represents different possible outputs, possibly from a class set C . Example datasets represent a small subset of example code distributed with the library." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "shares the same multi-headed attention core, there are significant differences between them including positional representations, masking, padding, and the use of sequence-to-sequence design. Additionally, various models are built to target different applications of NLP such as understanding, generation, and conditional generation, plus specialized use cases such as fast inference or multi-lingual applications.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 657.1114501953125, "coord_origin": "BOTTOMLEFT", "l": 71.24576568603516, "r": 292.0833740234375, "t": 776.1301879882812 }, "charspan": [ 0, 414 ], "page_no": 4 } ], "self_ref": "#/texts/26", "text": "shares the same multi-headed attention core, there are significant differences between them including positional representations, masking, padding, and the use of sequence-to-sequence design. Additionally, various models are built to target different applications of NLP such as understanding, generation, and conditional generation, plus specialized use cases such as fast inference or multi-lingual applications." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Practically, all models follow the same hierarchy of abstraction: a base class implements the model's computation graph from an encoding (projection on the embedding matrix) through the series of selfattention layers to the final encoder hidden states. The base class is specific to each model and closely follows the model's original implementation which gives users the flexibility to easily dissect the inner workings of each individual architecture. In most cases, each model is implemented in a single file to enable ease of extensibility.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 507.47357177734375, "coord_origin": "BOTTOMLEFT", "l": 71.01911926269531, "r": 292.175537109375, "t": 654.382080078125 }, "charspan": [ 0, 544 ], "page_no": 4 } ], "self_ref": "#/texts/27", "text": "Practically, all models follow the same hierarchy of abstraction: a base class implements the model's computation graph from an encoding (projection on the embedding matrix) through the series of selfattention layers to the final encoder hidden states. The base class is specific to each model and closely follows the model's original implementation which gives users the flexibility to easily dissect the inner workings of each individual architecture. In most cases, each model is implemented in a single file to enable ease of extensibility." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Wherever possible, different architectures follow the same API allowing users to switch easily between different models. A set of Auto classes provides a unified API that enables very fast switching between models and even between frameworks. These classes automatically instantiate with the configuration specified by the user-specified pretrained model.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 398.7656555175781, "coord_origin": "BOTTOMLEFT", "l": 71.00536346435547, "r": 292.179443359375, "t": 504.3974914550781 }, "charspan": [ 0, 355 ], "page_no": 4 } ], "self_ref": "#/texts/28", "text": "Wherever possible, different architectures follow the same API allowing users to switch easily between different models. A set of Auto classes provides a unified API that enables very fast switching between models and even between frameworks. These classes automatically instantiate with the configuration specified by the user-specified pretrained model." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Tokenizers A critical NLP-specific aspect of the library is the implementations of the tokenizers necessary to use each model. Tokenizer classes (each inheriting from a common base class) can either be instantiated from a corresponding pretrained model or can be configured manually. These classes store the vocabulary token-to-index map for their corresponding model and handle the encoding and decoding of input sequences according to a model's specific tokenization process. The tokenizers implemented are shown in Figure 2 (Right). Users can easily modify tokenizer with interfaces to add additional token mappings, special tokens (such as classification or separation tokens), or otherwise resize the vocabulary.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 185.94000244140625, "coord_origin": "BOTTOMLEFT", "l": 70.87411499023438, "r": 292.083251953125, "t": 386.3719787597656 }, "charspan": [ 0, 717 ], "page_no": 4 } ], "self_ref": "#/texts/29", "text": "Tokenizers A critical NLP-specific aspect of the library is the implementations of the tokenizers necessary to use each model. Tokenizer classes (each inheriting from a common base class) can either be instantiated from a corresponding pretrained model or can be configured manually. These classes store the vocabulary token-to-index map for their corresponding model and handle the encoding and decoding of input sequences according to a model's specific tokenization process. The tokenizers implemented are shown in Figure 2 (Right). Users can easily modify tokenizer with interfaces to add additional token mappings, special tokens (such as classification or separation tokens), or otherwise resize the vocabulary." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Tokenizers can also implement additional useful features for the users. These range from token type indices in the case of sequence classification to maximum length sequence truncating taking into account the added model-specific special tokens (most pretrained Transformer models have a maximum sequence length).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 90.32830810546875, "coord_origin": "BOTTOMLEFT", "l": 71.1646728515625, "r": 292.0804748535156, "t": 182.8790283203125 }, "charspan": [ 0, 313 ], "page_no": 4 } ], "self_ref": "#/texts/30", "text": "Tokenizers can also implement additional useful features for the users. These range from token type indices in the case of sequence classification to maximum length sequence truncating taking into account the added model-specific special tokens (most pretrained Transformer models have a maximum sequence length)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "For training on very large datasets, Python-based", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.6446533203125, "coord_origin": "BOTTOMLEFT", "l": 82.55223846435547, "r": 290.5442810058594, "t": 86.83929443359375 }, "charspan": [ 0, 49 ], "page_no": 4 } ], "self_ref": "#/texts/31", "text": "For training on very large datasets, Python-based" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "tokenization is often undesirably slow. In the most recent release, Transformers switched its implementation to use a highly-optimized tokenization library by default. This low-level library, available at https://github.com/huggingface/ tokenizers , is written in Rust to speed up the tokenization procedure both during training and deployment.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 670.2180786132812, "coord_origin": "BOTTOMLEFT", "l": 306.2556457519531, "r": 527.3560180664062, "t": 776.0233154296875 }, "charspan": [ 0, 344 ], "page_no": 4 } ], "self_ref": "#/texts/32", "text": "tokenization is often undesirably slow. In the most recent release, Transformers switched its implementation to use a highly-optimized tokenization library by default. This low-level library, available at https://github.com/huggingface/ tokenizers , is written in Rust to speed up the tokenization procedure both during training and deployment." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Heads Each Transformer can be paired with one out of several ready-implemented heads with outputs amenable to common types of tasks. These heads are implemented as additional wrapper classes on top of the base class, adding a specific output layer, and optional loss function, on top of the Transformer's contextual embeddings. The full set of implemented heads are shown in Figure 2 (Top). These classes follow a similar naming pattern: XXXForSequenceClassification where XXX is the name of the model and can be used for adaptation (fine-tuning) or pretraining. Some heads, such as conditional generation, support extra functionality like sampling and beam search.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 459.95550537109375, "coord_origin": "BOTTOMLEFT", "l": 306.0362854003906, "r": 527.4515380859375, "t": 660.9263916015625 }, "charspan": [ 0, 665 ], "page_no": 4 } ], "self_ref": "#/texts/33", "text": "Heads Each Transformer can be paired with one out of several ready-implemented heads with outputs amenable to common types of tasks. These heads are implemented as additional wrapper classes on top of the base class, adding a specific output layer, and optional loss function, on top of the Transformer's contextual embeddings. The full set of implemented heads are shown in Figure 2 (Top). These classes follow a similar naming pattern: XXXForSequenceClassification where XXX is the name of the model and can be used for adaptation (fine-tuning) or pretraining. Some heads, such as conditional generation, support extra functionality like sampling and beam search." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "For pretrained models, we release the heads used to pretrain the model itself. For instance, for BERT we release the language modeling and next sentence prediction heads which allows easy for adaptation using the pretraining objectives. We also make it easy for users to utilize the same core Transformer parameters with a variety of other heads for finetuning. While each head can be used generally, the library also includes a collection of examples that show each head on real problems. These examples demonstrate how a pretrained model can be adapted with a given head to achieve state-of-theart results on a large variety of NLP tasks.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 284.0336608886719, "coord_origin": "BOTTOMLEFT", "l": 306.4980163574219, "r": 527.359375, "t": 457.47381591796875 }, "charspan": [ 0, 640 ], "page_no": 4 } ], "self_ref": "#/texts/34", "text": "For pretrained models, we release the heads used to pretrain the model itself. For instance, for BERT we release the language modeling and next sentence prediction heads which allows easy for adaptation using the pretraining objectives. We also make it easy for users to utilize the same core Transformer parameters with a variety of other heads for finetuning. While each head can be used generally, the library also includes a collection of examples that show each head on real problems. These examples demonstrate how a pretrained model can be adapted with a given head to achieve state-of-theart results on a large variety of NLP tasks." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "4 Community Model Hub", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 259.47662353515625, "coord_origin": "BOTTOMLEFT", "l": 306.0682067871094, "r": 447.42681884765625, "t": 271.84405517578125 }, "charspan": [ 0, 21 ], "page_no": 4 } ], "self_ref": "#/texts/35", "text": "4 Community Model Hub" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Transformers aims to facilitate easy use and distribution of pretrained models. Inherently this is a community process; a single pretraining run facilitates fine-tuning on many specific tasks. The Model Hub makes it simple for any end-user to access a model for use with their own data. This hub now contains 2,097 user models, both pretrained and fine-tuned, from across the community. Figure 1 shows the increase and distribution of popular transformers over time. While core models like BERT and GPT-2 continue to be popular, other specialized models including DistilBERT (Sanh et al., 2019), which was developed for the library, are", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.671875, "coord_origin": "BOTTOMLEFT", "l": 306.1293640136719, "r": 527.359375, "t": 249.75677490234375 }, "charspan": [ 0, 636 ], "page_no": 4 } ], "self_ref": "#/texts/36", "text": "Transformers aims to facilitate easy use and distribution of pretrained models. Inherently this is a community process; a single pretraining run facilitates fine-tuning on many specific tasks. The Model Hub makes it simple for any end-user to access a model for use with their own data. This hub now contains 2,097 user models, both pretrained and fine-tuned, from across the community. Figure 1 shows the increase and distribution of popular transformers over time. While core models like BERT and GPT-2 continue to be popular, other specialized models including DistilBERT (Sanh et al., 2019), which was developed for the library, are" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "SciBERT", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 666.5567016601562, "coord_origin": "BOTTOMLEFT", "l": 88.66666412353516, "r": 140, "t": 680.5567016601562 }, "charspan": [ 0, 7 ], "page_no": 5 } ], "self_ref": "#/texts/37", "text": "SciBERT" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "This is the pretrained model presented in SciBERT: A Pretrained Language Model for Scientific Text, which is a BERT model trained on scientific text.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 638.5567016601562, "coord_origin": "BOTTOMLEFT", "l": 88.33333587646484, "r": 279, "t": 654.8900146484375 }, "charspan": [ 0, 149 ], "page_no": 5 } ], "self_ref": "#/texts/38", "text": "This is the pretrained model presented in SciBERT: A Pretrained Language Model for Scientific Text, which is a BERT model trained on scientific text." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The training corpus was papers taken from Semantic Scholar. Corpus size is 1.14M papers, 3.1B tokens. We use the full text of the papers in training; not just abstracts.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 606.2233276367188, "coord_origin": "BOTTOMLEFT", "l": 87.66666412353516, "r": 283, "t": 632.2233276367188 }, "charspan": [ 0, 169 ], "page_no": 5 } ], "self_ref": "#/texts/39", "text": "The training corpus was papers taken from Semantic Scholar. Corpus size is 1.14M papers, 3.1B tokens. We use the full text of the papers in training; not just abstracts." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "SciBERT has its own wordpiece vocabulary (scivocab) that's built to best match the training corpus. We trained cased and uncased versions.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 582.2233276367188, "coord_origin": "BOTTOMLEFT", "l": 88.33333587646484, "r": 275, "t": 600.8900146484375 }, "charspan": [ 0, 138 ], "page_no": 5 } ], "self_ref": "#/texts/40", "text": "SciBERT has its own wordpiece vocabulary (scivocab) that's built to best match the training corpus. We trained cased and uncased versions." }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 3: Transformers Model Hub. (Left) Example of a model page and model card for SciBERT (Beltagy et al., 2019), a pretrained model targeting extraction from scientific literature submitted by a community contributor. (Right) Example of an automatic inference widget for the pretrained BART (Lewis et al., 2019) model for summarization. Users can enter arbitrary text and a full version of the model is deployed on the fly to produce a summary.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 509.9199523925781, "coord_origin": "BOTTOMLEFT", "l": 71.2237777709961, "r": 527.2011108398438, "t": 568.3770751953125 }, "charspan": [ 0, 447 ], "page_no": 5 } ], "self_ref": "#/texts/41", "text": "Figure 3: Transformers Model Hub. (Left) Example of a model page and model card for SciBERT (Beltagy et al., 2019), a pretrained model targeting extraction from scientific literature submitted by a community contributor. (Right) Example of an automatic inference widget for the pretrained BART (Lewis et al., 2019) model for summarization. Users can enter arbitrary text and a full version of the model is deployed on the fly to produce a summary." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "now widely downloaded by the community.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 475.2874450683594, "coord_origin": "BOTTOMLEFT", "l": 71.67430877685547, "r": 262.85467529296875, "t": 486.22528076171875 }, "charspan": [ 0, 39 ], "page_no": 5 } ], "self_ref": "#/texts/42", "text": "now widely downloaded by the community." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "The user interface of the Model Hub is designed to be simple and open to the community. To upload a model, any user can sign up for an account and use a command-line interface to produce an archive consisting a tokenizer, transformer, and head. This bundle may be a model trained through the library or converted from a checkpoint of other popular training tools. These models are then stored and given a canonical name which a user can use to download, cache, and run the model either for finetuning or inference in two lines of code. To load FlauBERT (Le et al., 2020), a BERT model pretrained on a French training corpus, the command is:", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 284.9196472167969, "coord_origin": "BOTTOMLEFT", "l": 71.01830291748047, "r": 292.079833984375, "t": 471.972900390625 }, "charspan": [ 0, 640 ], "page_no": 5 } ], "self_ref": "#/texts/43", "text": "The user interface of the Model Hub is designed to be simple and open to the community. To upload a model, any user can sign up for an account and use a command-line interface to produce an archive consisting a tokenizer, transformer, and head. This bundle may be a model trained through the library or converted from a checkpoint of other popular training tools. These models are then stored and given a canonical name which a user can use to download, cache, and run the model either for finetuning or inference in two lines of code. To load FlauBERT (Le et al., 2020), a BERT model pretrained on a French training corpus, the command is:" }, { "children": [], "enumerated": null, "label": "code", "level": null, "marker": null, "orig": "1 tknzr = AutoTokenizer.from_pretrained( 2 \"flaubert/flaubert_base_uncased\") 3 model = AutoModel.from_pretrained( 4 \"flaubert/flaubert_base_uncased\")", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 235.93475341796875, "coord_origin": "BOTTOMLEFT", "l": 63.59010314941406, "r": 289.9710693359375, "t": 276.34600830078125 }, "charspan": [ 0, 149 ], "page_no": 5 } ], "self_ref": "#/texts/44", "text": "1 tknzr = AutoTokenizer.from_pretrained( 2 \"flaubert/flaubert_base_uncased\") 3 model = AutoModel.from_pretrained( 4 \"flaubert/flaubert_base_uncased\")" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "When a model is uploaded to the Model Hub, it is automatically given a landing page describing its core properties, architecture, and use cases. Additional model-specific metadata can be provided via a model card (Mitchell et al., 2018) that describes properties of its training, a citation to the work, datasets used during pretraining, and any caveats about known biases in the model and its predictions. An example model card is shown in Figure 3 (Left).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 104.59490966796875, "coord_origin": "BOTTOMLEFT", "l": 71.4018325805664, "r": 292.179443359375, "t": 223.946533203125 }, "charspan": [ 0, 457 ], "page_no": 5 } ], "self_ref": "#/texts/45", "text": "When a model is uploaded to the Model Hub, it is automatically given a landing page describing its core properties, architecture, and use cases. Additional model-specific metadata can be provided via a model card (Mitchell et al., 2018) that describes properties of its training, a citation to the work, datasets used during pretraining, and any caveats about known biases in the model and its predictions. An example model card is shown in Figure 3 (Left)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Since the Model Hub is specific to transformerbased models, we can target use cases that would", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.986328125, "coord_origin": "BOTTOMLEFT", "l": 71.52654266357422, "r": 292.078369140625, "t": 100.9810791015625 }, "charspan": [ 0, 94 ], "page_no": 5 } ], "self_ref": "#/texts/46", "text": "Since the Model Hub is specific to transformerbased models, we can target use cases that would" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "be difficult for more general model collections. For example, because each uploaded model includes metadata concerning its structure, the model page can include live inference that allows users to experiment with output of models on a real data. Figure 3 (Right) shows an example of the model page with live inference. Additionally, model pages include links to other model-specific tools like benchmarking and visualizations. For example, model pages can link to exBERT (Hoover et al., 2019), a Transformer visualization library.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 340.12896728515625, "coord_origin": "BOTTOMLEFT", "l": 306.1814880371094, "r": 527.3591918945312, "t": 485.967529296875 }, "charspan": [ 0, 530 ], "page_no": 5 } ], "self_ref": "#/texts/47", "text": "be difficult for more general model collections. For example, because each uploaded model includes metadata concerning its structure, the model page can include live inference that allows users to experiment with output of models on a real data. Figure 3 (Right) shows an example of the model page with live inference. Additionally, model pages include links to other model-specific tools like benchmarking and visualizations. For example, model pages can link to exBERT (Hoover et al., 2019), a Transformer visualization library." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Community Case Studies The Model Hub highlights how Transformers is used by a variety of different community stakeholders. We summarize three specific observed use-cases in practice. We highlight specific systems developed by users with different goals following the architect, trainer, and end-user distinction of Strobelt et al. (2017):", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 230.20263671875, "coord_origin": "BOTTOMLEFT", "l": 306.3488464355469, "r": 527.3536987304688, "t": 322.404052734375 }, "charspan": [ 0, 338 ], "page_no": 5 } ], "self_ref": "#/texts/48", "text": "Community Case Studies The Model Hub highlights how Transformers is used by a variety of different community stakeholders. We summarize three specific observed use-cases in practice. We highlight specific systems developed by users with different goals following the architect, trainer, and end-user distinction of Strobelt et al. (2017):" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Case 1: Model Architects AllenAI, a major NLP research lab, developed a new pretrained model for improved extraction from biomedical texts called SciBERT (Beltagy et al., 2019). They were able to train the model utilizing data from PubMed to produce a masked language model with state-ofthe-art results on targeted text. They then used the Model Hub to distribute the model and promote it as part of their CORD - COVID-19 challenge, making it trivial for the community to use.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 92.1527099609375, "coord_origin": "BOTTOMLEFT", "l": 306.3707275390625, "r": 527.3514404296875, "t": 224.97540283203125 }, "charspan": [ 0, 476 ], "page_no": 5 } ], "self_ref": "#/texts/49", "text": "Case 1: Model Architects AllenAI, a major NLP research lab, developed a new pretrained model for improved extraction from biomedical texts called SciBERT (Beltagy et al., 2019). They were able to train the model utilizing data from PubMed to produce a masked language model with state-ofthe-art results on targeted text. They then used the Model Hub to distribute the model and promote it as part of their CORD - COVID-19 challenge, making it trivial for the community to use." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Case 2: Task Trainers Researchers at NYU were", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 76.56763458251953, "coord_origin": "BOTTOMLEFT", "l": 306.4453430175781, "r": 525.5425415039062, "t": 86.90167236328125 }, "charspan": [ 0, 45 ], "page_no": 5 } ], "self_ref": "#/texts/50", "text": "Case 2: Task Trainers Researchers at NYU were" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "interested in developing a test bed for the performance of Transformers on a variety of different semantic recognition tasks. Their framework Jiant (Pruksachatkun et al., 2020) allows them to experiment with different ways of pretraining models and comparing their outputs. They used the Transformers API as a generic front-end and performed fine-tuning on a variety of different models, leading to research on the structure of BERT (Tenney et al., 2019).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 643.7344970703125, "coord_origin": "BOTTOMLEFT", "l": 71.02261352539062, "r": 292.083251953125, "t": 775.955078125 }, "charspan": [ 0, 455 ], "page_no": 6 } ], "self_ref": "#/texts/51", "text": "interested in developing a test bed for the performance of Transformers on a variety of different semantic recognition tasks. Their framework Jiant (Pruksachatkun et al., 2020) allows them to experiment with different ways of pretraining models and comparing their outputs. They used the Transformers API as a generic front-end and performed fine-tuning on a variety of different models, leading to research on the structure of BERT (Tenney et al., 2019)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Case 3: Application Users Plot.ly, a company focused on user dashboards and analytics, was interested in deploying a model for automatic document summarization. They wanted an approach that scaled well and was simple to deploy, but had no need to train or fine-tune the model. They were able to search the Model Hub and find DistilBART , a pretrained and fine-tuned summarization model designed for accurate, fast inference. They were able to run and deploy the model directly from the hub with no required research or ML expertise.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 493.51177978515625, "coord_origin": "BOTTOMLEFT", "l": 71.08113861083984, "r": 292.07537841796875, "t": 640.251708984375 }, "charspan": [ 0, 532 ], "page_no": 6 } ], "self_ref": "#/texts/52", "text": "Case 3: Application Users Plot.ly, a company focused on user dashboards and analytics, was interested in deploying a model for automatic document summarization. They wanted an approach that scaled well and was simple to deploy, but had no need to train or fine-tune the model. They were able to search the Model Hub and find DistilBART , a pretrained and fine-tuned summarization model designed for accurate, fast inference. They were able to run and deploy the model directly from the hub with no required research or ML expertise." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "5 Deployment", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 466.9922180175781, "coord_origin": "BOTTOMLEFT", "l": 71.33447265625, "r": 151.693359375, "t": 479.0706481933594 }, "charspan": [ 0, 12 ], "page_no": 6 } ], "self_ref": "#/texts/53", "text": "5 Deployment" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "An increasingly important goal of Transformers is to make it easy to efficiently deploy model to production. Different users have different production needs, and deployment often requires solving significantly different challenges than training. The library thereforce allows for several different strategies for production deployment.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 362.8888854980469, "coord_origin": "BOTTOMLEFT", "l": 71.21505737304688, "r": 292.0832214355469, "t": 455.3365173339844 }, "charspan": [ 0, 335 ], "page_no": 6 } ], "self_ref": "#/texts/54", "text": "An increasingly important goal of Transformers is to make it easy to efficiently deploy model to production. Different users have different production needs, and deployment often requires solving significantly different challenges than training. The library thereforce allows for several different strategies for production deployment." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "One core propery of the libary is that models are available both in PyTorch and TensorFlow, and there is interoperability between both frameworks. A model trained in one of frameworks can be saved through standard serialization and be reloaded from the saved files in the other framework seamlessly. This makes it particularly easy to switch from one framework to the other one along the model lifetime (training, serving, etc.).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 240.43072509765625, "coord_origin": "BOTTOMLEFT", "l": 71.15364837646484, "r": 292.1783752441406, "t": 359.832763671875 }, "charspan": [ 0, 429 ], "page_no": 6 } ], "self_ref": "#/texts/55", "text": "One core propery of the libary is that models are available both in PyTorch and TensorFlow, and there is interoperability between both frameworks. A model trained in one of frameworks can be saved through standard serialization and be reloaded from the saved files in the other framework seamlessly. This makes it particularly easy to switch from one framework to the other one along the model lifetime (training, serving, etc.)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Each framework has deployment recommendations. For example, in PyTorch, models are compatible with TorchScript, an intermediate representation of a PyTorch model that can then be run either in Python in a more efficient way, or in a highperformance environment such as C++. Fine-tuned models can thus be exported to production-friendly environment, and run through TorchServing. TensorFlow includes several serving options within its ecosystem, and these can be used directly.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 103.9495849609375, "coord_origin": "BOTTOMLEFT", "l": 71.39894104003906, "r": 292.0832214355469, "t": 236.77777099609375 }, "charspan": [ 0, 476 ], "page_no": 6 } ], "self_ref": "#/texts/56", "text": "Each framework has deployment recommendations. For example, in PyTorch, models are compatible with TorchScript, an intermediate representation of a PyTorch model that can then be run either in Python in a more efficient way, or in a highperformance environment such as C++. Fine-tuned models can thus be exported to production-friendly environment, and run through TorchServing. TensorFlow includes several serving options within its ecosystem, and these can be used directly." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Transformers can also export models to intermediate neural network formats for further compila-", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.80743408203125, "coord_origin": "BOTTOMLEFT", "l": 71.23482513427734, "r": 292.0760803222656, "t": 100.8079833984375 }, "charspan": [ 0, 95 ], "page_no": 6 } ], "self_ref": "#/texts/57", "text": "Transformers can also export models to intermediate neural network formats for further compila-" }, { "children": [], "enumerated": null, "label": "caption", "level": null, "marker": null, "orig": "Figure 4: Experiments with Transformers inference in collaboration with ONNX.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 615.0130615234375, "coord_origin": "BOTTOMLEFT", "l": 306.52197265625, "r": 525.5469970703125, "t": 636.6781616210938 }, "charspan": [ 0, 77 ], "page_no": 6 } ], "self_ref": "#/texts/58", "text": "Figure 4: Experiments with Transformers inference in collaboration with ONNX." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "tion. It supports converting models to the Open Neural Network Exchange format (ONNX) for deployment. Not only does this allow the model to be run in a standardized interoperable format, but also leads to significant speed-ups. Figure 4 shows experiments run in collaboration with the ONNX team to optimize BERT, RoBERTa, and GPT-2 from the Transformers library. Using this intermediate format, ONNX was able to achieve nearly a 4x speedup on this model. The team is also experimenting with other promising intermediate formats such as JAX/XLA (Bradbury et al., 2018) and TVM (Chen et al., 2018).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 416.2196350097656, "coord_origin": "BOTTOMLEFT", "l": 306.23046875, "r": 527.3594970703125, "t": 589.6441040039062 }, "charspan": [ 0, 596 ], "page_no": 6 } ], "self_ref": "#/texts/59", "text": "tion. It supports converting models to the Open Neural Network Exchange format (ONNX) for deployment. Not only does this allow the model to be run in a standardized interoperable format, but also leads to significant speed-ups. Figure 4 shows experiments run in collaboration with the ONNX team to optimize BERT, RoBERTa, and GPT-2 from the Transformers library. Using this intermediate format, ONNX was able to achieve nearly a 4x speedup on this model. The team is also experimenting with other promising intermediate formats such as JAX/XLA (Bradbury et al., 2018) and TVM (Chen et al., 2018)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Finally, as Transformers become more widely used in all NLP applications, it is increasingly important to deploy to edge devices such as phones or home electronics. Models can use adapters to convert models to CoreML weights that are suitable to be embedded inside a iOS application, to enable on-the-edge machine learning. Code is also made available$^{3}$. Similar methods can be used for Android devices.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 293.51263427734375, "coord_origin": "BOTTOMLEFT", "l": 306.228515625, "r": 527.3536376953125, "t": 412.32318115234375 }, "charspan": [ 0, 407 ], "page_no": 6 } ], "self_ref": "#/texts/60", "text": "Finally, as Transformers become more widely used in all NLP applications, it is increasingly important to deploy to edge devices such as phones or home electronics. Models can use adapters to convert models to CoreML weights that are suitable to be embedded inside a iOS application, to enable on-the-edge machine learning. Code is also made available$^{3}$. Similar methods can be used for Android devices." }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "6 Conclusion", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 266.9573669433594, "coord_origin": "BOTTOMLEFT", "l": 306.5513610839844, "r": 382.34271240234375, "t": 278.31634521484375 }, "charspan": [ 0, 12 ], "page_no": 6 } ], "self_ref": "#/texts/61", "text": "6 Conclusion" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "As Transformer and pretraining play larger roles in NLP, it is important for these models to be accessible to researchers and end-users. Transformers is an open-source library and community designed to facilitate users to access large-scale pretrained models, to build and experiment on top of them, and to deploy them in downstream tasks with stateof-the-art performance. Transformers has gained significant organic traction since its release and is set up to continue to provide core infrastructure while helping to facilitate access to new models.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 108.4984130859375, "coord_origin": "BOTTOMLEFT", "l": 306.1824951171875, "r": 527.3589477539062, "t": 254.68658447265625 }, "charspan": [ 0, 550 ], "page_no": 6 } ], "self_ref": "#/texts/62", "text": "As Transformer and pretraining play larger roles in NLP, it is important for these models to be accessible to researchers and end-users. Transformers is an open-source library and community designed to facilitate users to access large-scale pretrained models, to build and experiment on top of them, and to deploy them in downstream tasks with stateof-the-art performance. Transformers has gained significant organic traction since its release and is set up to continue to provide core infrastructure while helping to facilitate access to new models." }, { "children": [], "enumerated": null, "label": "footnote", "level": null, "marker": null, "orig": "$^{3}$https://github.com/huggingface/", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 86.205078125, "coord_origin": "BOTTOMLEFT", "l": 319.1553649902344, "r": 490.1900634765625, "t": 96.09271240234375 }, "charspan": [ 0, 37 ], "page_no": 6 } ], "self_ref": "#/texts/63", "text": "$^{3}$https://github.com/huggingface/" }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "swift-coreml-transformers", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 77.60594177246094, "coord_origin": "BOTTOMLEFT", "l": 307.2760009765625, "r": 442.7672424316406, "t": 84.52799987792969 }, "charspan": [ 0, 25 ], "page_no": 6 } ], "self_ref": "#/texts/64", "text": "swift-coreml-transformers" }, { "children": [], "enumerated": null, "label": "section_header", "level": 1, "marker": null, "orig": "References", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 765.6243286132812, "coord_origin": "BOTTOMLEFT", "l": 71.65863037109375, "r": 127.74907684326172, "t": 776.8631591796875 }, "charspan": [ 0, 10 ], "page_no": 7 } ], "self_ref": "#/texts/65", "text": "References" }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "a. PyTorch Hub. https://pytorch.org/hub/ . Accessed: 2020-6-29.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 737.1751098632812, "coord_origin": "BOTTOMLEFT", "l": 71.17964172363281, "r": 291.924072265625, "t": 757.851318359375 }, "charspan": [ 0, 63 ], "page_no": 7 } ], "self_ref": "#/texts/66", "text": "a. PyTorch Hub. https://pytorch.org/hub/ . Accessed: 2020-6-29." }, { "children": [], "enumerated": false, "label": "list_item", "level": null, "marker": "-", "orig": "b. TensorFlow hub. https://www.tensorflow. org/hub . Accessed: 2020-6-29.", "parent": { "$ref": "#/groups/0" }, "prov": [ { "bbox": { "b": 705.765380859375, "coord_origin": "BOTTOMLEFT", "l": 71.41734313964844, "r": 294.0343322753906, "t": 726.62939453125 }, "charspan": [ 0, 73 ], "page_no": 7 } ], "self_ref": "#/texts/67", "text": "b. TensorFlow hub. https://www.tensorflow. org/hub . Accessed: 2020-6-29." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Alan Akbik, Tanja Bergmann, Duncan Blythe, Kashif Rasul, Stefan Schweter, and Roland Vollgraf. 2019. Flair: An easy-to-use framework for state-of-the-art nlp. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics (Demonstrations) , pages 5459. aclweb.org.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 619.50146484375, "coord_origin": "BOTTOMLEFT", "l": 71.31597137451172, "r": 292.0140380859375, "t": 695.657470703125 }, "charspan": [ 0, 318 ], "page_no": 7 } ], "self_ref": "#/texts/68", "text": "Alan Akbik, Tanja Bergmann, Duncan Blythe, Kashif Rasul, Stefan Schweter, and Roland Vollgraf. 2019. Flair: An easy-to-use framework for state-of-the-art nlp. In Proceedings of the 2019 Conference of the North American Chapter of the Association for Computational Linguistics (Demonstrations) , pages 5459. aclweb.org." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Iz Beltagy, Kyle Lo, and Arman Cohan. 2019. SciBERT: A pretrained language model for scientific text. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) , pages 36153620.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 534.1513061523438, "coord_origin": "BOTTOMLEFT", "l": 71.14820098876953, "r": 292.0140075683594, "t": 609.40576171875 }, "charspan": [ 0, 298 ], "page_no": 7 } ], "self_ref": "#/texts/69", "text": "Iz Beltagy, Kyle Lo, and Arman Cohan. 2019. SciBERT: A pretrained language model for scientific text. In Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP) , pages 36153620." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Iz Beltagy, Matthew E Peters, and Arman Cohan. 2020. Longformer: The Long-Document transformer.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 502.8877258300781, "coord_origin": "BOTTOMLEFT", "l": 71.35470581054688, "r": 292.0140686035156, "t": 523.4490966796875 }, "charspan": [ 0, 95 ], "page_no": 7 } ], "self_ref": "#/texts/70", "text": "Iz Beltagy, Matthew E Peters, and Arman Cohan. 2020. Longformer: The Long-Document transformer." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Antoine Bosselut, Hannah Rashkin, Maarten Sap, Chaitanya Malaviya, Asli elikyilmaz, and Yejin Choi. 2019. Comet: Commonsense transformers for automatic knowledge graph construction. In ACL .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 449.5399169921875, "coord_origin": "BOTTOMLEFT", "l": 70.98684692382812, "r": 292.0140075683594, "t": 492.69488525390625 }, "charspan": [ 0, 190 ], "page_no": 7 } ], "self_ref": "#/texts/71", "text": "Antoine Bosselut, Hannah Rashkin, Maarten Sap, Chaitanya Malaviya, Asli elikyilmaz, and Yejin Choi. 2019. Comet: Commonsense transformers for automatic knowledge graph construction. In ACL ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "James Bradbury, Roy Frostig, Peter Hawkins, Matthew James Johnson, Chris Leary, Dougal Maclaurin, and Skye Wanderman-Milne. 2018. JAX: composable transformations of Python+NumPy programs.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 385.11328125, "coord_origin": "BOTTOMLEFT", "l": 70.93028259277344, "r": 292.0140075683594, "t": 439.15185546875 }, "charspan": [ 0, 187 ], "page_no": 7 } ], "self_ref": "#/texts/72", "text": "James Bradbury, Roy Frostig, Peter Hawkins, Matthew James Johnson, Chris Leary, Dougal Maclaurin, and Skye Wanderman-Milne. 2018. JAX: composable transformations of Python+NumPy programs." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Tianqi Chen, Thierry Moreau, Ziheng Jiang, Lianmin Zheng, Eddie Yan, Haichen Shen, Meghan Cowan, Leyuan Wang, Yuwei Hu, Luis Ceze, et al. 2018. { TVM } : An automated end-to-end optimizing compiler for deep learning. In 13th { USENIX } Symposium on Operating Systems Design and Implementation ( { OSDI } 18) , pages 578-594.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 299.34588623046875, "coord_origin": "BOTTOMLEFT", "l": 71.27062225341797, "r": 292.01397705078125, "t": 375.197265625 }, "charspan": [ 0, 324 ], "page_no": 7 } ], "self_ref": "#/texts/73", "text": "Tianqi Chen, Thierry Moreau, Ziheng Jiang, Lianmin Zheng, Eddie Yan, Haichen Shen, Meghan Cowan, Leyuan Wang, Yuwei Hu, Luis Ceze, et al. 2018. { TVM } : An automated end-to-end optimizing compiler for deep learning. In 13th { USENIX } Symposium on Operating Systems Design and Implementation ( { OSDI } 18) , pages 578-594." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Kevin Clark, Minh-Thang Luong, Quoc V Le, and Christopher D Manning. 2020. ELECTRA: Pretraining text encoders as discriminators rather than generators.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 246.54156494140625, "coord_origin": "BOTTOMLEFT", "l": 71.53278350830078, "r": 291.9243469238281, "t": 289.3968505859375 }, "charspan": [ 0, 151 ], "page_no": 7 } ], "self_ref": "#/texts/74", "text": "Kevin Clark, Minh-Thang Luong, Quoc V Le, and Christopher D Manning. 2020. ELECTRA: Pretraining text encoders as discriminators rather than generators." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Zihang Dai, Zhilin Yang, Yiming Yang, Jaime Carbonell, Quoc V Le, and Ruslan Salakhutdinov. 2019. Transformer-XL: Attentive language models beyond a Fixed-Length context.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 193.19769287109375, "coord_origin": "BOTTOMLEFT", "l": 70.75564575195312, "r": 292.0140380859375, "t": 236.427001953125 }, "charspan": [ 0, 170 ], "page_no": 7 } ], "self_ref": "#/texts/75", "text": "Zihang Dai, Zhilin Yang, Yiming Yang, Jaime Carbonell, Quoc V Le, and Ruslan Salakhutdinov. 2019. Transformer-XL: Attentive language models beyond a Fixed-Length context." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. BERT: Pre-training of deep bidirectional transformers for language understanding.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 139.9075927734375, "coord_origin": "BOTTOMLEFT", "l": 71.08364868164062, "r": 291.92437744140625, "t": 183.38958740234375 }, "charspan": [ 0, 153 ], "page_no": 7 } ], "self_ref": "#/texts/76", "text": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. BERT: Pre-training of deep bidirectional transformers for language understanding." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Matt Gardner, Joel Grus, Mark Neumann, Oyvind Tafjord, Pradeep Dasigi, Nelson Liu, Matthew Peters, Michael Schmitz, and Luke Zettlemoyer. 2018. AllenNLP: A deep semantic natural language processing platform.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 76.27606201171875, "coord_origin": "BOTTOMLEFT", "l": 71.01158905029297, "r": 292.0140380859375, "t": 130.1431884765625 }, "charspan": [ 0, 207 ], "page_no": 7 } ], "self_ref": "#/texts/77", "text": "Matt Gardner, Joel Grus, Mark Neumann, Oyvind Tafjord, Pradeep Dasigi, Nelson Liu, Matthew Peters, Michael Schmitz, and Luke Zettlemoyer. 2018. AllenNLP: A deep semantic natural language processing platform." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Matthew Honnibal and Ines Montani. 2017. spacy 2: Natural language understanding with bloom embeddings, convolutional neural networks and incremental parsing. To appear , 7(1).", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 732.5965576171875, "coord_origin": "BOTTOMLEFT", "l": 306.07757568359375, "r": 527.2003784179688, "t": 775.7552490234375 }, "charspan": [ 0, 176 ], "page_no": 7 } ], "self_ref": "#/texts/78", "text": "Matthew Honnibal and Ines Montani. 2017. spacy 2: Natural language understanding with bloom embeddings, convolutional neural networks and incremental parsing. To appear , 7(1)." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Benjamin Hoover, Hendrik Strobelt, and Sebastian Gehrmann. 2019. exBERT: A visual analysis tool to explore learned representations in transformers models.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 680.2274780273438, "coord_origin": "BOTTOMLEFT", "l": 306.6997985839844, "r": 527.2003173828125, "t": 722.7184448242188 }, "charspan": [ 0, 154 ], "page_no": 7 } ], "self_ref": "#/texts/79", "text": "Benjamin Hoover, Hendrik Strobelt, and Sebastian Gehrmann. 2019. exBERT: A visual analysis tool to explore learned representations in transformers models." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Jeremy Howard and Sebastian Ruder. 2018. Universal language model fine-tuning for text classification. In ACL .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 638.3195190429688, "coord_origin": "BOTTOMLEFT", "l": 306.57659912109375, "r": 525.5465698242188, "t": 670.1641235351562 }, "charspan": [ 0, 111 ], "page_no": 7 } ], "self_ref": "#/texts/80", "text": "Jeremy Howard and Sebastian Ruder. 2018. Universal language model fine-tuning for text classification. In ACL ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Zhiting Hu, Haoran Shi, Bowen Tan, Wentao Wang, Zichao Yang, Tiancheng Zhao, Junxian He, Lianhui Qin, Di Wang, Xuezhe Ma, Zhengzhong Liu, Xiaodan Liang, Wangrong Zhu, Devendra Singh Sachan, and Eric P Xing. 2018. Texar: A modularized, versatile, and extensible toolkit for text generation.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 563.5346069335938, "coord_origin": "BOTTOMLEFT", "l": 306.35015869140625, "r": 527.2003173828125, "t": 628.191650390625 }, "charspan": [ 0, 289 ], "page_no": 7 } ], "self_ref": "#/texts/81", "text": "Zhiting Hu, Haoran Shi, Bowen Tan, Wentao Wang, Zichao Yang, Tiancheng Zhao, Junxian He, Lianhui Qin, Di Wang, Xuezhe Ma, Zhengzhong Liu, Xiaodan Liang, Wangrong Zhu, Devendra Singh Sachan, and Eric P Xing. 2018. Texar: A modularized, versatile, and extensible toolkit for text generation." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Mandar Joshi, Danqi Chen, Yinhan Liu, Daniel S Weld, Luke Zettlemoyer, and Omer Levy. 2019. Spanbert: Improving pre-training by representing and predicting spans. arXiv preprint arXiv:1907.10529 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 510.1098937988281, "coord_origin": "BOTTOMLEFT", "l": 306.5328063964844, "r": 527.2002563476562, "t": 553.4983520507812 }, "charspan": [ 0, 196 ], "page_no": 7 } ], "self_ref": "#/texts/82", "text": "Mandar Joshi, Danqi Chen, Yinhan Liu, Daniel S Weld, Luke Zettlemoyer, and Omer Levy. 2019. Spanbert: Improving pre-training by representing and predicting spans. arXiv preprint arXiv:1907.10529 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Marcin Junczys-Dowmunt, Roman Grundkiewicz, Tomasz Dwojak, Hieu Hoang, Kenneth Heafield, Tom Neckermann, Frank Seide, Ulrich Germann, Alham Fikri Aji, Nikolay Bogoychev, Andr'e F T Martins, and Alexandra Birch. 2018. Marian: Fast neural machine translation in c++.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 435.88275146484375, "coord_origin": "BOTTOMLEFT", "l": 306.544677734375, "r": 527.2007446289062, "t": 500.22540283203125 }, "charspan": [ 0, 264 ], "page_no": 7 } ], "self_ref": "#/texts/83", "text": "Marcin Junczys-Dowmunt, Roman Grundkiewicz, Tomasz Dwojak, Hieu Hoang, Kenneth Heafield, Tom Neckermann, Frank Seide, Ulrich Germann, Alham Fikri Aji, Nikolay Bogoychev, Andr'e F T Martins, and Alexandra Birch. 2018. Marian: Fast neural machine translation in c++." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Douwe Kiela, Suvrat Bhooshan, Hamed Firooz, and Davide Testuggine. 2019. Supervised multimodal bitransformers for classifying images and text.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 393.5155029296875, "coord_origin": "BOTTOMLEFT", "l": 306.68902587890625, "r": 525.546630859375, "t": 425.3257141113281 }, "charspan": [ 0, 142 ], "page_no": 7 } ], "self_ref": "#/texts/84", "text": "Douwe Kiela, Suvrat Bhooshan, Hamed Firooz, and Davide Testuggine. 2019. Supervised multimodal bitransformers for classifying images and text." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Nikita Kitaev, Łukasz Kaiser, and Anselm Levskaya. 2020. Reformer: The efficient transformer.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 363.02581787109375, "coord_origin": "BOTTOMLEFT", "l": 306.3611755371094, "r": 527.2900390625, "t": 383.43798828125 }, "charspan": [ 0, 93 ], "page_no": 7 } ], "self_ref": "#/texts/85", "text": "Nikita Kitaev, Łukasz Kaiser, and Anselm Levskaya. 2020. Reformer: The efficient transformer." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Guillaume Klein, Yoon Kim, Yuntian Deng, Jean Senellart, and Alexander Rush. 2017. OpenNMT: Opensource toolkit for neural machine translation. In Proceedings of ACL 2017, System Demonstrations , pages 67-72, Vancouver, Canada. Association for Computational Linguistics.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 287.96917724609375, "coord_origin": "BOTTOMLEFT", "l": 306.509765625, "r": 527.200439453125, "t": 352.8445129394531 }, "charspan": [ 0, 269 ], "page_no": 7 } ], "self_ref": "#/texts/86", "text": "Guillaume Klein, Yoon Kim, Yuntian Deng, Jean Senellart, and Alexander Rush. 2017. OpenNMT: Opensource toolkit for neural machine translation. In Proceedings of ACL 2017, System Demonstrations , pages 67-72, Vancouver, Canada. Association for Computational Linguistics." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Guillaume Lample and Alexis Conneau. 2019a. Crosslingual language model pretraining. In NeurIPS .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 256.80889892578125, "coord_origin": "BOTTOMLEFT", "l": 306.2541809082031, "r": 527.2002563476562, "t": 278.16156005859375 }, "charspan": [ 0, 97 ], "page_no": 7 } ], "self_ref": "#/texts/87", "text": "Guillaume Lample and Alexis Conneau. 2019a. Crosslingual language model pretraining. In NeurIPS ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Guillaume Lample and Alexis Conneau. 2019b. Crosslingual language model pretraining.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 226.18426513671875, "coord_origin": "BOTTOMLEFT", "l": 306.1673278808594, "r": 527.2003173828125, "t": 246.6683349609375 }, "charspan": [ 0, 84 ], "page_no": 7 } ], "self_ref": "#/texts/88", "text": "Guillaume Lample and Alexis Conneau. 2019b. Crosslingual language model pretraining." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, and Radu Soricut. 2019. ALBERT: A lite BERT for self-supervised learning of language representations.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 173.06494140625, "coord_origin": "BOTTOMLEFT", "l": 306.2449645996094, "r": 527.2899780273438, "t": 216.12347412109375 }, "charspan": [ 0, 177 ], "page_no": 7 } ], "self_ref": "#/texts/89", "text": "Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin Gimpel, Piyush Sharma, and Radu Soricut. 2019. ALBERT: A lite BERT for self-supervised learning of language representations." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hang Le, Lo¨ıc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoˆıt Crabb'e, Laurent Besacier, and Didier Schwab. 2020. Flaubert: Unsupervised language model pre-training for french. In Proceedings of The 12th Language Resources and Evaluation Conference , pages 2479-2490, Marseille, France. European Language Resources Association.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 75.8814697265625, "coord_origin": "BOTTOMLEFT", "l": 307.0087890625, "r": 527.2003173828125, "t": 162.6871337890625 }, "charspan": [ 0, 381 ], "page_no": 7 } ], "self_ref": "#/texts/90", "text": "Hang Le, Lo¨ıc Vial, Jibril Frej, Vincent Segonne, Maximin Coavoux, Benjamin Lecouteux, Alexandre Allauzen, Benoˆıt Crabb'e, Laurent Besacier, and Didier Schwab. 2020. Flaubert: Unsupervised language model pre-training for french. In Proceedings of The 12th Language Resources and Evaluation Conference , pages 2479-2490, Marseille, France. European Language Resources Association." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov, and Luke Zettlemoyer. 2019. BART: Denoising Sequence-to-Sequence pretraining for natural language generation, translation, and comprehension.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 710.797119140625, "coord_origin": "BOTTOMLEFT", "l": 71.00279998779297, "r": 292.0140075683594, "t": 775.480712890625 }, "charspan": [ 0, 246 ], "page_no": 8 } ], "self_ref": "#/texts/91", "text": "Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoyanov, and Luke Zettlemoyer. 2019. BART: Denoising Sequence-to-Sequence pretraining for natural language generation, translation, and comprehension." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019a. RoBERTa: A robustly optimized BERT pretraining approach.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 643.5848388671875, "coord_origin": "BOTTOMLEFT", "l": 71.29876708984375, "r": 292.01397705078125, "t": 698.0454711914062 }, "charspan": [ 0, 199 ], "page_no": 8 } ], "self_ref": "#/texts/92", "text": "Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. 2019a. RoBERTa: A robustly optimized BERT pretraining approach." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar S. Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke S. Zettlemoyer, and Veselin Stoyanov. 2019b. Roberta: A robustly optimized bert pretraining approach. ArXiv , abs/1907.11692.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 577.03515625, "coord_origin": "BOTTOMLEFT", "l": 71.15786743164062, "r": 292.01397705078125, "t": 631.5889282226562 }, "charspan": [ 0, 229 ], "page_no": 8 } ], "self_ref": "#/texts/93", "text": "Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar S. Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke S. Zettlemoyer, and Veselin Stoyanov. 2019b. Roberta: A robustly optimized bert pretraining approach. ArXiv , abs/1907.11692." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Edward Loper and Steven Bird. 2002. NLTK: The natural language toolkit.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 543.4266357421875, "coord_origin": "BOTTOMLEFT", "l": 71.4495620727539, "r": 291.9244079589844, "t": 564.6576538085938 }, "charspan": [ 0, 71 ], "page_no": 8 } ], "self_ref": "#/texts/94", "text": "Edward Loper and Steven Bird. 2002. NLTK: The natural language toolkit." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Christopher D Manning, Mihai Surdeanu, John Bauer, Jenny Rose Finkel, Steven Bethard, and David McClosky. 2014. The stanford CoreNLP natural language processing toolkit. In Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations , pages 55-60. aclweb.org.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 454.76593017578125, "coord_origin": "BOTTOMLEFT", "l": 71.30555725097656, "r": 292.0086975097656, "t": 530.6755981445312 }, "charspan": [ 0, 306 ], "page_no": 8 } ], "self_ref": "#/texts/95", "text": "Christopher D Manning, Mihai Surdeanu, John Bauer, Jenny Rose Finkel, Steven Bethard, and David McClosky. 2014. The stanford CoreNLP natural language processing toolkit. In Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations , pages 55-60. aclweb.org." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Bryan McCann, James Bradbury, Caiming Xiong, and Richard Socher. 2017. Learned in translation: Contextualized word vectors. In I Guyon, U V Luxburg, S Bengio, H Wallach, R Fergus, S Vishwanathan, and R Garnett, editors, Advances in Neural Information Processing Systems 30 , pages 6294-6305. Curran Associates, Inc.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 366.57733154296875, "coord_origin": "BOTTOMLEFT", "l": 71.5431900024414, "r": 291.92437744140625, "t": 442.0590515136719 }, "charspan": [ 0, 315 ], "page_no": 8 } ], "self_ref": "#/texts/96", "text": "Bryan McCann, James Bradbury, Caiming Xiong, and Richard Socher. 2017. Learned in translation: Contextualized word vectors. In I Guyon, U V Luxburg, S Bengio, H Wallach, R Fergus, S Vishwanathan, and R Garnett, editors, Advances in Neural Information Processing Systems 30 , pages 6294-6305. Curran Associates, Inc." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, Inioluwa Deborah Raji, and Timnit Gebru. 2018. Model cards for model reporting.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 310.808349609375, "coord_origin": "BOTTOMLEFT", "l": 71.47232055664062, "r": 291.5159606933594, "t": 353.32666015625 }, "charspan": [ 0, 188 ], "page_no": 8 } ], "self_ref": "#/texts/97", "text": "Margaret Mitchell, Simone Wu, Andrew Zaldivar, Parker Barnes, Lucy Vasserman, Ben Hutchinson, Elena Spitzer, Inioluwa Deborah Raji, and Timnit Gebru. 2018. Model cards for model reporting." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Myle Ott, Sergey Edunov, Alexei Baevski, Angela Fan, Sam Gross, Nathan Ng, David Grangier, and Michael Auli. 2019. fairseq: A fast, extensible toolkit for sequence modeling.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 254.59356689453125, "coord_origin": "BOTTOMLEFT", "l": 71.04486846923828, "r": 290.27056884765625, "t": 297.34014892578125 }, "charspan": [ 0, 173 ], "page_no": 8 } ], "self_ref": "#/texts/98", "text": "Myle Ott, Sergey Edunov, Alexei Baevski, Angela Fan, Sam Gross, Nathan Ng, David Grangier, and Michael Auli. 2019. fairseq: A fast, extensible toolkit for sequence modeling." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Matthew E Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 199.27037048339844, "coord_origin": "BOTTOMLEFT", "l": 71.00860595703125, "r": 291.9243469238281, "t": 241.59246826171875 }, "charspan": [ 0, 159 ], "page_no": 8 } ], "self_ref": "#/texts/99", "text": "Matthew E Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Yada Pruksachatkun, Phil Yeres, Haokun Liu, Jason Phang, Phu Mon Htut, Alex Wang, Ian Tenney, and Samuel R Bowman. 2020. jiant: A software toolkit for research on general-purpose text understanding models. arXiv preprint arXiv:2003.02249 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 132.36328125, "coord_origin": "BOTTOMLEFT", "l": 71.41844177246094, "r": 290.2706298828125, "t": 185.8568115234375 }, "charspan": [ 0, 239 ], "page_no": 8 } ], "self_ref": "#/texts/100", "text": "Yada Pruksachatkun, Phil Yeres, Haokun Liu, Jason Phang, Phu Mon Htut, Alex Wang, Ian Tenney, and Samuel R Bowman. 2020. jiant: A software toolkit for research on general-purpose text understanding models. arXiv preprint arXiv:2003.02249 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Peng Qi, Yuhao Zhang, Yuhui Zhang, Jason Bolton, and Christopher D Manning. 2020. Stanza: A python natural language processing toolkit for many human languages.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 76.17864990234375, "coord_origin": "BOTTOMLEFT", "l": 71.46034240722656, "r": 291.51593017578125, "t": 119.47564697265625 }, "charspan": [ 0, 160 ], "page_no": 8 } ], "self_ref": "#/texts/101", "text": "Peng Qi, Yuhao Zhang, Yuhui Zhang, Jason Bolton, and Christopher D Manning. 2020. Stanza: A python natural language processing toolkit for many human languages." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. OpenAI Blog , 1(8):9.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 733.0943603515625, "coord_origin": "BOTTOMLEFT", "l": 306.483642578125, "r": 526.7919311523438, "t": 775.425537109375 }, "charspan": [ 0, 165 ], "page_no": 8 } ], "self_ref": "#/texts/102", "text": "Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners. OpenAI Blog , 1(8):9." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2019. Exploring the limits of transfer learning with a unified Text-to-Text transformer.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 669.3334350585938, "coord_origin": "BOTTOMLEFT", "l": 306.3033142089844, "r": 527.2003784179688, "t": 722.6685791015625 }, "charspan": [ 0, 213 ], "page_no": 8 } ], "self_ref": "#/texts/103", "text": "Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. 2019. Exploring the limits of transfer learning with a unified Text-to-Text transformer." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Victor Sanh, Lysandre Debut, Julien Chaumond, and Thomas Wolf. 2019. DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 627.490478515625, "coord_origin": "BOTTOMLEFT", "l": 306.5908203125, "r": 525.5465698242188, "t": 658.7568969726562 }, "charspan": [ 0, 147 ], "page_no": 8 } ], "self_ref": "#/texts/104", "text": "Victor Sanh, Lysandre Debut, Julien Chaumond, and Thomas Wolf. 2019. DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, and Bryan Catanzaro. 2019. Megatron-lm: Training multi-billion parameter language models using gpu model parallelism. arXiv preprint arXiv:1909.08053 .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 563.440673828125, "coord_origin": "BOTTOMLEFT", "l": 306.4382629394531, "r": 527.2003784179688, "t": 617.4739379882812 }, "charspan": [ 0, 230 ], "page_no": 8 } ], "self_ref": "#/texts/105", "text": "Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, and Bryan Catanzaro. 2019. Megatron-lm: Training multi-billion parameter language models using gpu model parallelism. arXiv preprint arXiv:1909.08053 ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Hendrik Strobelt, Sebastian Gehrmann, Hanspeter Pfister, and Alexander M Rush. 2017. Lstmvis: A tool for visual analysis of hidden state dynamics in recurrent neural networks. IEEE transactions on visualization and computer graphics , 24(1):667-676.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 499.582763671875, "coord_origin": "BOTTOMLEFT", "l": 306.7300720214844, "r": 527.200439453125, "t": 553.271484375 }, "charspan": [ 0, 249 ], "page_no": 8 } ], "self_ref": "#/texts/106", "text": "Hendrik Strobelt, Sebastian Gehrmann, Hanspeter Pfister, and Alexander M Rush. 2017. Lstmvis: A tool for visual analysis of hidden state dynamics in recurrent neural networks. IEEE transactions on visualization and computer graphics , 24(1):667-676." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ian Tenney, Dipanjan Das, and Ellie Pavlick. 2019. Bert rediscovers the classical nlp pipeline. In ACL .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 468.90106201171875, "coord_origin": "BOTTOMLEFT", "l": 306.7170715332031, "r": 527.2900390625, "t": 489.2518310546875 }, "charspan": [ 0, 104 ], "page_no": 8 } ], "self_ref": "#/texts/107", "text": "Ian Tenney, Dipanjan Das, and Ellie Pavlick. 2019. Bert rediscovers the classical nlp pipeline. In ACL ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ashish Vaswani, Samy Bengio, Eugene Brevdo, Francois Chollet, Aidan N. Gomez, Stephan Gouws, Llion Jones, Łukasz Kaiser, Nal Kalchbrenner, Niki Parmar, Ryan Sepassi, Noam Shazeer, and Jakob Uszkoreit. 2018. Tensor2tensor for neural machine translation. CoRR , abs/1803.07416.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 394.36474609375, "coord_origin": "BOTTOMLEFT", "l": 306.4836120605469, "r": 527.2005004882812, "t": 458.455322265625 }, "charspan": [ 0, 275 ], "page_no": 8 } ], "self_ref": "#/texts/108", "text": "Ashish Vaswani, Samy Bengio, Eugene Brevdo, Francois Chollet, Aidan N. Gomez, Stephan Gouws, Llion Jones, Łukasz Kaiser, Nal Kalchbrenner, Niki Parmar, Ryan Sepassi, Noam Shazeer, and Jakob Uszkoreit. 2018. Tensor2tensor for neural machine translation. CoRR , abs/1803.07416." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł Ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In I Guyon, U V Luxburg, S Bengio, H Wallach, R Fergus, S Vishwanathan, and R Garnett, editors, Advances in Neural Information Processing Systems 30 , pages 5998-6008. Curran Associates, Inc.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 297.72784423828125, "coord_origin": "BOTTOMLEFT", "l": 306.4910888671875, "r": 527.2008056640625, "t": 384.1600646972656 }, "charspan": [ 0, 350 ], "page_no": 8 } ], "self_ref": "#/texts/109", "text": "Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Ł Ukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In I Guyon, U V Luxburg, S Bengio, H Wallach, R Fergus, S Vishwanathan, and R Garnett, editors, Advances in Neural Information Processing Systems 30 , pages 5998-6008. Curran Associates, Inc." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Alex Wang, Yada Pruksachatkun, Nikita Nangia, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel R. Bowman. 2019. Superglue: A stickier benchmark for general-purpose language understanding systems. ArXiv , abs/1905.00537.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 233.91778564453125, "coord_origin": "BOTTOMLEFT", "l": 306.388671875, "r": 527.200439453125, "t": 287.34796142578125 }, "charspan": [ 0, 237 ], "page_no": 8 } ], "self_ref": "#/texts/110", "text": "Alex Wang, Yada Pruksachatkun, Nikita Nangia, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel R. Bowman. 2019. Superglue: A stickier benchmark for general-purpose language understanding systems. ArXiv , abs/1905.00537." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Alex Wang, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel R. Bowman. 2018. Glue: A multi-task benchmark and analysis platform for natural language understanding. In ICLR .", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 180.75543212890625, "coord_origin": "BOTTOMLEFT", "l": 306.4911804199219, "r": 527.2900390625, "t": 223.44586181640625 }, "charspan": [ 0, 191 ], "page_no": 8 } ], "self_ref": "#/texts/111", "text": "Alex Wang, Amanpreet Singh, Julian Michael, Felix Hill, Omer Levy, and Samuel R. Bowman. 2018. Glue: A multi-task benchmark and analysis platform for natural language understanding. In ICLR ." }, { "children": [], "enumerated": null, "label": "text", "level": null, "marker": null, "orig": "Zhilin Yang, Zihang Dai, Yiming Yang, Jaime G. Carbonell, Ruslan Salakhutdinov, and Quoc V. Le. 2019. Xlnet: Generalized autoregressive pretraining for language understanding. ArXiv , abs/1906.08237.", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 128.247314453125, "coord_origin": "BOTTOMLEFT", "l": 306.262939453125, "r": 527.2900390625, "t": 170.82635498046875 }, "charspan": [ 0, 199 ], "page_no": 8 } ], "self_ref": "#/texts/112", "text": "Zhilin Yang, Zihang Dai, Yiming Yang, Jaime G. Carbonell, Ruslan Salakhutdinov, and Quoc V. Le. 2019. Xlnet: Generalized autoregressive pretraining for language understanding. ArXiv , abs/1906.08237." } ]

[ { "annotations": [], "captions": [], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 750.6070556640625, "coord_origin": "BOTTOMLEFT", "l": 94.95348358154297, "r": 120.46149444580078, "t": 774.6229248046875 }, "charspan": [ 0, 0 ], "page_no": 1 } ], "references": [], "self_ref": "#/pictures/0" }, { "annotations": [], "captions": [ { "$ref": "#/texts/15" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 636.427734375, "coord_origin": "BOTTOMLEFT", "l": 66.24333953857422, "r": 536, "t": 739.33349609375 }, "charspan": [ 0, 105 ], "page_no": 2 } ], "references": [], "self_ref": "#/pictures/1" }, { "annotations": [], "captions": [ { "$ref": "#/texts/25" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 393.6780090332031, "coord_origin": "BOTTOMLEFT", "l": 345.091796875, "r": 497.1534118652344, "t": 568.6835327148438 }, "charspan": [ 0, 858 ], "page_no": 3 } ], "references": [], "self_ref": "#/pictures/2" }, { "annotations": [], "captions": [], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 697.90087890625, "coord_origin": "BOTTOMLEFT", "l": 86.91616821289062, "r": 285.61199951171875, "t": 778.6142578125 }, "charspan": [ 0, 0 ], "page_no": 5 } ], "references": [], "self_ref": "#/pictures/3" }, { "annotations": [], "captions": [ { "$ref": "#/texts/41" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 581.1311645507812, "coord_origin": "BOTTOMLEFT", "l": 316.5115966796875, "r": 511.6044616699219, "t": 773.5567016601562 }, "charspan": [ 0, 447 ], "page_no": 5 } ], "references": [], "self_ref": "#/pictures/4" }, { "annotations": [], "captions": [ { "$ref": "#/texts/58" } ], "children": [], "footnotes": [], "label": "picture", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 647.065673828125, "coord_origin": "BOTTOMLEFT", "l": 306.4806823730469, "r": 525.6688842773438, "t": 779.533935546875 }, "charspan": [ 0, 77 ], "page_no": 6 } ], "references": [], "self_ref": "#/pictures/5" } ]

[ { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 81.49700927734375, "r": 107.54794311523438, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Name" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 211.3480224609375, "r": 234.37713623046875, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Input" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 276.028076171875, "r": 313.319580078125, "t": 773.1983032226562 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Heads Output" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 333.6717529296875, "r": 358.24993896484375, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "Tasks" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 431.0027160644531, "r": 486.1918640136719, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "Ex. Datasets" } ], [ { "bbox": { "b": 730.7396240234375, "coord_origin": "BOTTOMLEFT", "l": 81.49700164794922, "r": 169.66433715820312, "t": 740.4923706054688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Language Modeling" }, { "bbox": { "b": 730.9796142578125, "coord_origin": "BOTTOMLEFT", "l": 208.2270050048828, "r": 237.0075225830078, "t": 740.6669311523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "x$_{1:}$$_{n}$$_{-}$$_{1}$" }, { "bbox": { "b": 730.9796142578125, "coord_origin": "BOTTOMLEFT", "l": 274.78802490234375, "r": 306.680908203125, "t": 741.27783203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "x$_{n}$ ∈ V" }, { "bbox": { "b": 730.7396240234375, "coord_origin": "BOTTOMLEFT", "l": 334.0119934082031, "r": 382.4811096191406, "t": 740.4923706054688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "Generation" }, { "bbox": { "b": 730.7396240234375, "coord_origin": "BOTTOMLEFT", "l": 430.4920654296875, "r": 490.9393615722656, "t": 740.4923706054688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "WikiText-103" } ], [ { "bbox": { "b": 717.1906127929688, "coord_origin": "BOTTOMLEFT", "l": 81.49699401855469, "r": 185.427978515625, "t": 726.943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Sequence Classification" }, { "bbox": { "b": 717.4306030273438, "coord_origin": "BOTTOMLEFT", "l": 212.42098999023438, "r": 231.9641876220703, "t": 727.117919921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "x$_{1:}$$_{N}$" }, { "bbox": { "b": 717.4306030273438, "coord_origin": "BOTTOMLEFT", "l": 278.4549865722656, "r": 303.2744445800781, "t": 727.7288208007812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "y ∈ C" }, { "bbox": { "b": 717.1906127929688, "coord_origin": "BOTTOMLEFT", "l": 334.011962890625, "r": 396.1392822265625, "t": 726.943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "Classification," }, { "bbox": { "b": 717.1906127929688, "coord_origin": "BOTTOMLEFT", "l": 431.0069885253906, "r": 486.2615661621094, "t": 726.943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "GLUE, SST," } ], [ { "bbox": { "b": 690.0916137695312, "coord_origin": "BOTTOMLEFT", "l": 81.49697875976562, "r": 171.48614501953125, "t": 699.8443603515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Question Answering" }, { "bbox": { "b": 690.3316040039062, "coord_origin": "BOTTOMLEFT", "l": 197.38198852539062, "r": 247.003173828125, "t": 700.0189208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "x$_{1:}$$_{M}$,x$_{M}$$_{:}$$_{N}$" }, { "bbox": { "b": 690.0916137695312, "coord_origin": "BOTTOMLEFT", "l": 260.30596923828125, "r": 322.0574951171875, "t": 700.0189208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "y span [1 : N ]" }, { "bbox": { "b": 676.5426025390625, "coord_origin": "BOTTOMLEFT", "l": 334.0119323730469, "r": 419.05059814453125, "t": 713.3933715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "Sentiment Analysis QA, Reading Comprehension" }, { "bbox": { "b": 649.4446411132812, "coord_origin": "BOTTOMLEFT", "l": 431.0049743652344, "r": 510.08697509765625, "t": 713.3933715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "MNLI SQuAD, Natural Questions SWAG, ARC" } ], [ { "bbox": { "b": 622.1986083984375, "coord_origin": "BOTTOMLEFT", "l": 81.49691772460938, "r": 184.2170867919922, "t": 672.746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Token Classification Multiple Choice Masked LM Conditional Generation" }, { "bbox": { "b": 622.4385986328125, "coord_origin": "BOTTOMLEFT", "l": 205.30799865722656, "r": 238.8274688720703, "t": 672.9208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "x$_{1:}$$_{N}$ x$_{1:}$$_{N}$, X x$_{1:}$$_{N}$$_{\\}$$_{n}$ x$_{1:}$$_{N}$" }, { "bbox": { "b": 622.4385986328125, "coord_origin": "BOTTOMLEFT", "l": 265.2760009765625, "r": 315.7386779785156, "t": 674.8402099609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "y$_{1:}$$_{N}$ ∈ C N y ∈ X x$_{n}$ ∈ V y$_{1:}$$_{M}$ ∈ V M" }, { "bbox": { "b": 622.1986083984375, "coord_origin": "BOTTOMLEFT", "l": 333.6739807128906, "r": 396.7719421386719, "t": 672.746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "NER, Tagging Text Selection Pretraining Translation," }, { "bbox": { "b": 622.1986083984375, "coord_origin": "BOTTOMLEFT", "l": 430.4920654296875, "r": 516.0307006835938, "t": 672.746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "OntoNotes, WNUT Wikitext, C4 WMT, IWSLT," } ] ], "num_cols": 5, "num_rows": 5, "table_cells": [ { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 81.49700927734375, "r": 107.54794311523438, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Name" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 211.3480224609375, "r": 234.37713623046875, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 0, "text": "Input" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 276.028076171875, "r": 313.319580078125, "t": 773.1983032226562 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 3, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 0, "text": "Heads Output" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 333.6717529296875, "r": 358.24993896484375, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 4, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 0, "text": "Tasks" }, { "bbox": { "b": 749.765625, "coord_origin": "BOTTOMLEFT", "l": 431.0027160644531, "r": 486.1918640136719, "t": 759.5183715820312 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 5, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 0, "text": "Ex. Datasets" }, { "bbox": { "b": 730.7396240234375, "coord_origin": "BOTTOMLEFT", "l": 81.49700164794922, "r": 169.66433715820312, "t": 740.4923706054688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Language Modeling" }, { "bbox": { "b": 730.9796142578125, "coord_origin": "BOTTOMLEFT", "l": 208.2270050048828, "r": 237.0075225830078, "t": 740.6669311523438 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "x$_{1:}$$_{n}$$_{-}$$_{1}$" }, { "bbox": { "b": 730.9796142578125, "coord_origin": "BOTTOMLEFT", "l": 274.78802490234375, "r": 306.680908203125, "t": 741.27783203125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 1, "text": "x$_{n}$ ∈ V" }, { "bbox": { "b": 730.7396240234375, "coord_origin": "BOTTOMLEFT", "l": 334.0119934082031, "r": 382.4811096191406, "t": 740.4923706054688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 1, "text": "Generation" }, { "bbox": { "b": 730.7396240234375, "coord_origin": "BOTTOMLEFT", "l": 430.4920654296875, "r": 490.9393615722656, "t": 740.4923706054688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 1, "text": "WikiText-103" }, { "bbox": { "b": 717.1906127929688, "coord_origin": "BOTTOMLEFT", "l": 81.49699401855469, "r": 185.427978515625, "t": 726.943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Sequence Classification" }, { "bbox": { "b": 717.4306030273438, "coord_origin": "BOTTOMLEFT", "l": 212.42098999023438, "r": 231.9641876220703, "t": 727.117919921875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "x$_{1:}$$_{N}$" }, { "bbox": { "b": 717.4306030273438, "coord_origin": "BOTTOMLEFT", "l": 278.4549865722656, "r": 303.2744445800781, "t": 727.7288208007812 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 2, "text": "y ∈ C" }, { "bbox": { "b": 717.1906127929688, "coord_origin": "BOTTOMLEFT", "l": 334.011962890625, "r": 396.1392822265625, "t": 726.943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 2, "text": "Classification," }, { "bbox": { "b": 717.1906127929688, "coord_origin": "BOTTOMLEFT", "l": 431.0069885253906, "r": 486.2615661621094, "t": 726.943359375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 2, "text": "GLUE, SST," }, { "bbox": { "b": 690.0916137695312, "coord_origin": "BOTTOMLEFT", "l": 81.49697875976562, "r": 171.48614501953125, "t": 699.8443603515625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Question Answering" }, { "bbox": { "b": 690.3316040039062, "coord_origin": "BOTTOMLEFT", "l": 197.38198852539062, "r": 247.003173828125, "t": 700.0189208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "x$_{1:}$$_{M}$,x$_{M}$$_{:}$$_{N}$" }, { "bbox": { "b": 690.0916137695312, "coord_origin": "BOTTOMLEFT", "l": 260.30596923828125, "r": 322.0574951171875, "t": 700.0189208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 3, "text": "y span [1 : N ]" }, { "bbox": { "b": 676.5426025390625, "coord_origin": "BOTTOMLEFT", "l": 334.0119323730469, "r": 419.05059814453125, "t": 713.3933715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 3, "text": "Sentiment Analysis QA, Reading Comprehension" }, { "bbox": { "b": 649.4446411132812, "coord_origin": "BOTTOMLEFT", "l": 431.0049743652344, "r": 510.08697509765625, "t": 713.3933715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 3, "text": "MNLI SQuAD, Natural Questions SWAG, ARC" }, { "bbox": { "b": 622.1986083984375, "coord_origin": "BOTTOMLEFT", "l": 81.49691772460938, "r": 184.2170867919922, "t": 672.746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "Token Classification Multiple Choice Masked LM Conditional Generation" }, { "bbox": { "b": 622.4385986328125, "coord_origin": "BOTTOMLEFT", "l": 205.30799865722656, "r": 238.8274688720703, "t": 672.9208984375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "x$_{1:}$$_{N}$ x$_{1:}$$_{N}$, X x$_{1:}$$_{N}$$_{\\}$$_{n}$ x$_{1:}$$_{N}$" }, { "bbox": { "b": 622.4385986328125, "coord_origin": "BOTTOMLEFT", "l": 265.2760009765625, "r": 315.7386779785156, "t": 674.8402099609375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 3, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 2, "start_row_offset_idx": 4, "text": "y$_{1:}$$_{N}$ ∈ C N y ∈ X x$_{n}$ ∈ V y$_{1:}$$_{M}$ ∈ V M" }, { "bbox": { "b": 622.1986083984375, "coord_origin": "BOTTOMLEFT", "l": 333.6739807128906, "r": 396.7719421386719, "t": 672.746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 4, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 3, "start_row_offset_idx": 4, "text": "NER, Tagging Text Selection Pretraining Translation," }, { "bbox": { "b": 622.1986083984375, "coord_origin": "BOTTOMLEFT", "l": 430.4920654296875, "r": 516.0307006835938, "t": 672.746337890625 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 5, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 4, "start_row_offset_idx": 4, "text": "OntoNotes, WNUT Wikitext, C4 WMT, IWSLT," } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 603.1106567382812, "coord_origin": "BOTTOMLEFT", "l": 74.91996765136719, "r": 522.26025390625, "t": 780.163818359375 }, "charspan": [ 0, 0 ], "page_no": 3 } ], "references": [], "self_ref": "#/tables/0" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 552.2249755859375, "coord_origin": "BOTTOMLEFT", "l": 154.3350067138672, "r": 218.08779907226562, "t": 562.0322875976562 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Transformers" }, { "bbox": { "b": 552.2249755859375, "coord_origin": "BOTTOMLEFT", "l": 154.3350067138672, "r": 218.08779907226562, "t": 562.0322875976562 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Transformers" } ], [ { "bbox": { "b": 533.1216430664062, "coord_origin": "BOTTOMLEFT", "l": 135.05099487304688, "r": 237.37054443359375, "t": 543.6598510742188 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Masked [ x$_{1:}$$_{N}$$_{\\}$$_{n}$ ⇒ x$_{n}$ ]" }, { "bbox": { "b": 533.1216430664062, "coord_origin": "BOTTOMLEFT", "l": 135.05099487304688, "r": 237.37054443359375, "t": 543.6598510742188 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Masked [ x$_{1:}$$_{N}$$_{\\}$$_{n}$ ⇒ x$_{n}$ ]" } ], [ { "bbox": { "b": 500.3986511230469, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 121.90894317626953, "t": 523.7003784179688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "BERT RoBERTa" }, { "bbox": { "b": 500.3986511230469, "coord_origin": "BOTTOMLEFT", "l": 206.55264282226562, "r": 294.447265625, "t": 523.7003784179688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "(Devlin et al., 2018) (Liu et al., 2019a)" } ], [ { "bbox": { "b": 481.3726501464844, "coord_origin": "BOTTOMLEFT", "l": 120.03600311279297, "r": 252.3855438232422, "t": 491.91082763671875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Autoregressive [ x$_{1:}$$_{n}$$_{-}$$_{1}$ ⇒ x$_{n}$ ]" }, { "bbox": { "b": 481.3726501464844, "coord_origin": "BOTTOMLEFT", "l": 120.03600311279297, "r": 252.3855438232422, "t": 491.91082763671875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Autoregressive [ x$_{1:}$$_{n}$$_{-}$$_{1}$ ⇒ x$_{n}$ ]" } ], [ { "bbox": { "b": 435.24761962890625, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 135.76348876953125, "t": 472.0983581542969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "GPT / GPT-2 Trans-XL XLNet" }, { "bbox": { "b": 435.24761962890625, "coord_origin": "BOTTOMLEFT", "l": 200.22535705566406, "r": 294.447265625, "t": 472.0983581542969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "(Radford et al., 2019) (Dai et al., 2019) (Yang et al., 2019)" } ], [ { "bbox": { "b": 416.2216491699219, "coord_origin": "BOTTOMLEFT", "l": 122.44499969482422, "r": 249.97654724121094, "t": 426.75982666015625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": true, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Seq-to-Seq [ ∼ x$_{1:}$$_{N}$ ⇒ x$_{1:}$$_{N}$ ]" }, { "bbox": { "b": 416.2216491699219, "coord_origin": "BOTTOMLEFT", "l": 122.44499969482422, "r": 249.97654724121094, "t": 426.75982666015625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": true, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Seq-to-Seq [ ∼ x$_{1:}$$_{N}$ ⇒ x$_{1:}$$_{N}$ ]" } ], [ { "bbox": { "b": 383.6456298828125, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 106.03619384765625, "t": 406.9473571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "BART T5" }, { "bbox": { "b": 383.6456298828125, "coord_origin": "BOTTOMLEFT", "l": 208.98538208007812, "r": 294.447265625, "t": 406.9473571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "(Lewis et al., 2019) (Raffel et al., 2019)" } ], [ { "bbox": { "b": 351.06964111328125, "coord_origin": "BOTTOMLEFT", "l": 136.9409942626953, "r": 235.48289489746094, "t": 360.8223571777344 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Specialty: Multimodal" }, { "bbox": { "b": 351.06964111328125, "coord_origin": "BOTTOMLEFT", "l": 136.9409942626953, "r": 235.48289489746094, "t": 360.8223571777344 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Specialty: Multimodal" } ], [ { "bbox": { "b": 332.04364013671875, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 111.31620788574219, "t": 341.7963562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "MMBT" }, { "bbox": { "b": 313.01763916015625, "coord_origin": "BOTTOMLEFT", "l": 130.28599548339844, "r": 294.447265625, "t": 341.7963562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "(Kiela et al., 2019) Specialty: Long-Distance" } ], [ { "bbox": { "b": 215.29063415527344, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 129.45803833007812, "t": 252.141357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "ALBERT Electra DistilBERT" }, { "bbox": { "b": 215.29063415527344, "coord_origin": "BOTTOMLEFT", "l": 211.73446655273438, "r": 294.447265625, "t": 252.141357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "(Lan et al., 2019) (Clark et al., 2020) (Sanh et al., 2019)" } ], [ { "bbox": { "b": 196.2636260986328, "coord_origin": "BOTTOMLEFT", "l": 135.4239959716797, "r": 236.99862670898438, "t": 206.016357421875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Specialty: Multilingual" }, { "bbox": { "b": 196.2636260986328, "coord_origin": "BOTTOMLEFT", "l": 135.4239959716797, "r": 236.99862670898438, "t": 206.016357421875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Specialty: Multilingual" } ], [ { "bbox": { "b": 177.2376251220703, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 149.18170166015625, "t": 186.9903564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "XLM/RoBERTa" }, { "bbox": { "b": 177.2376251220703, "coord_origin": "BOTTOMLEFT", "l": 161.13807678222656, "r": 294.447265625, "t": 186.9903564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "(Lample and Conneau, 2019b)" } ] ], "num_cols": 2, "num_rows": 12, "table_cells": [ { "bbox": { "b": 552.2249755859375, "coord_origin": "BOTTOMLEFT", "l": 154.3350067138672, "r": 218.08779907226562, "t": 562.0322875976562 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Transformers" }, { "bbox": { "b": 533.1216430664062, "coord_origin": "BOTTOMLEFT", "l": 135.05099487304688, "r": 237.37054443359375, "t": 543.6598510742188 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Masked [ x$_{1:}$$_{N}$$_{\\}$$_{n}$ ⇒ x$_{n}$ ]" }, { "bbox": { "b": 500.3986511230469, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 121.90894317626953, "t": 523.7003784179688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "BERT RoBERTa" }, { "bbox": { "b": 500.3986511230469, "coord_origin": "BOTTOMLEFT", "l": 206.55264282226562, "r": 294.447265625, "t": 523.7003784179688 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "(Devlin et al., 2018) (Liu et al., 2019a)" }, { "bbox": { "b": 481.3726501464844, "coord_origin": "BOTTOMLEFT", "l": 120.03600311279297, "r": 252.3855438232422, "t": 491.91082763671875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Autoregressive [ x$_{1:}$$_{n}$$_{-}$$_{1}$ ⇒ x$_{n}$ ]" }, { "bbox": { "b": 435.24761962890625, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 135.76348876953125, "t": 472.0983581542969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "GPT / GPT-2 Trans-XL XLNet" }, { "bbox": { "b": 435.24761962890625, "coord_origin": "BOTTOMLEFT", "l": 200.22535705566406, "r": 294.447265625, "t": 472.0983581542969 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "(Radford et al., 2019) (Dai et al., 2019) (Yang et al., 2019)" }, { "bbox": { "b": 416.2216491699219, "coord_origin": "BOTTOMLEFT", "l": 122.44499969482422, "r": 249.97654724121094, "t": 426.75982666015625 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": true, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "Seq-to-Seq [ ∼ x$_{1:}$$_{N}$ ⇒ x$_{1:}$$_{N}$ ]" }, { "bbox": { "b": 383.6456298828125, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 106.03619384765625, "t": 406.9473571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "BART T5" }, { "bbox": { "b": 383.6456298828125, "coord_origin": "BOTTOMLEFT", "l": 208.98538208007812, "r": 294.447265625, "t": 406.9473571777344 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "(Lewis et al., 2019) (Raffel et al., 2019)" }, { "bbox": { "b": 351.06964111328125, "coord_origin": "BOTTOMLEFT", "l": 136.9409942626953, "r": 235.48289489746094, "t": 360.8223571777344 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Specialty: Multimodal" }, { "bbox": { "b": 332.04364013671875, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 111.31620788574219, "t": 341.7963562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "MMBT" }, { "bbox": { "b": 313.01763916015625, "coord_origin": "BOTTOMLEFT", "l": 130.28599548339844, "r": 294.447265625, "t": 341.7963562011719 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "(Kiela et al., 2019) Specialty: Long-Distance" }, { "bbox": { "b": 215.29063415527344, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 129.45803833007812, "t": 252.141357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 9, "text": "ALBERT Electra DistilBERT" }, { "bbox": { "b": 215.29063415527344, "coord_origin": "BOTTOMLEFT", "l": 211.73446655273438, "r": 294.447265625, "t": 252.141357421875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 10, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 9, "text": "(Lan et al., 2019) (Clark et al., 2020) (Sanh et al., 2019)" }, { "bbox": { "b": 196.2636260986328, "coord_origin": "BOTTOMLEFT", "l": 135.4239959716797, "r": 236.99862670898438, "t": 206.016357421875 }, "col_span": 2, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 11, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 10, "text": "Specialty: Multilingual" }, { "bbox": { "b": 177.2376251220703, "coord_origin": "BOTTOMLEFT", "l": 77.97799682617188, "r": 149.18170166015625, "t": 186.9903564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 11, "text": "XLM/RoBERTa" }, { "bbox": { "b": 177.2376251220703, "coord_origin": "BOTTOMLEFT", "l": 161.13807678222656, "r": 294.447265625, "t": 186.9903564453125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 12, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 11, "text": "(Lample and Conneau, 2019b)" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 173.25628662109375, "coord_origin": "BOTTOMLEFT", "l": 71.48153686523438, "r": 300.6355895996094, "t": 568.4135131835938 }, "charspan": [ 0, 0 ], "page_no": 3 } ], "references": [], "self_ref": "#/tables/1" }, { "captions": [], "children": [], "data": { "grid": [ [ { "bbox": { "b": 322.58599853515625, "coord_origin": "BOTTOMLEFT", "l": 391.2099914550781, "r": 441.61004638671875, "t": 332.3932800292969 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Tokenizers" }, { "bbox": { "b": 322.58599853515625, "coord_origin": "BOTTOMLEFT", "l": 391.2099914550781, "r": 441.61004638671875, "t": 332.3932800292969 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Tokenizers" } ], [ { "bbox": { "b": 303.483642578125, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 365.2579345703125, "t": 313.2363586425781 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Name" }, { "bbox": { "b": 303.483642578125, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 483.55621337890625, "t": 313.2363586425781 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Ex. Uses" } ], [ { "bbox": { "b": 284.4566345214844, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 432.1634216308594, "t": 294.2093505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Character-Level BPE" }, { "bbox": { "b": 284.4566345214844, "coord_origin": "BOTTOMLEFT", "l": 444.1197814941406, "r": 493.6143493652344, "t": 294.2093505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "NMT, GPT" } ], [ { "bbox": { "b": 270.9076232910156, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 410.5852355957031, "t": 280.66033935546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Byte-Level BPE" }, { "bbox": { "b": 270.9076232910156, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 472.8107604980469, "t": 280.66033935546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "GPT-2" } ], [ { "bbox": { "b": 257.3586120605469, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 386.8033752441406, "t": 267.111328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "WordPiece" }, { "bbox": { "b": 257.3586120605469, "coord_origin": "BOTTOMLEFT", "l": 444.1197814941406, "r": 471.3489074707031, "t": 267.111328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "BERT" } ], [ { "bbox": { "b": 243.80960083007812, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 402.21795654296875, "t": 253.5623321533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "SentencePiece" }, { "bbox": { "b": 243.80960083007812, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 474.41436767578125, "t": 253.5623321533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "XLNet" } ], [ { "bbox": { "b": 230.26060485839844, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 377.9888610839844, "t": 240.01333618164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Unigram" }, { "bbox": { "b": 230.26060485839844, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 460.48345947265625, "t": 240.01333618164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "LM" } ], [ { "bbox": { "b": 216.71160888671875, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 381.61065673828125, "t": 226.46434020996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Character" }, { "bbox": { "b": 216.71160888671875, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 485.9234924316406, "t": 226.46434020996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Reformer" } ], [ { "bbox": { "b": 203.16160583496094, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 373.1561279296875, "t": 212.91433715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Custom" }, { "bbox": { "b": 203.16160583496094, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 489.5780334472656, "t": 212.91433715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Bio-Chem" } ] ], "num_cols": 2, "num_rows": 9, "table_cells": [ { "bbox": { "b": 322.58599853515625, "coord_origin": "BOTTOMLEFT", "l": 391.2099914550781, "r": 441.61004638671875, "t": 332.3932800292969 }, "col_span": 2, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 1, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 0, "text": "Tokenizers" }, { "bbox": { "b": 303.483642578125, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 365.2579345703125, "t": 313.2363586425781 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 1, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 1, "text": "Name" }, { "bbox": { "b": 303.483642578125, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 483.55621337890625, "t": 313.2363586425781 }, "col_span": 1, "column_header": true, "end_col_offset_idx": 2, "end_row_offset_idx": 2, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 1, "text": "Ex. Uses" }, { "bbox": { "b": 284.4566345214844, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 432.1634216308594, "t": 294.2093505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 2, "text": "Character-Level BPE" }, { "bbox": { "b": 284.4566345214844, "coord_origin": "BOTTOMLEFT", "l": 444.1197814941406, "r": 493.6143493652344, "t": 294.2093505859375 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 3, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 2, "text": "NMT, GPT" }, { "bbox": { "b": 270.9076232910156, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 410.5852355957031, "t": 280.66033935546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 3, "text": "Byte-Level BPE" }, { "bbox": { "b": 270.9076232910156, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 472.8107604980469, "t": 280.66033935546875 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 4, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 3, "text": "GPT-2" }, { "bbox": { "b": 257.3586120605469, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 386.8033752441406, "t": 267.111328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 4, "text": "WordPiece" }, { "bbox": { "b": 257.3586120605469, "coord_origin": "BOTTOMLEFT", "l": 444.1197814941406, "r": 471.3489074707031, "t": 267.111328125 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 5, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 4, "text": "BERT" }, { "bbox": { "b": 243.80960083007812, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 402.21795654296875, "t": 253.5623321533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 5, "text": "SentencePiece" }, { "bbox": { "b": 243.80960083007812, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 474.41436767578125, "t": 253.5623321533203 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 6, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 5, "text": "XLNet" }, { "bbox": { "b": 230.26060485839844, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 377.9888610839844, "t": 240.01333618164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 6, "text": "Unigram" }, { "bbox": { "b": 230.26060485839844, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 460.48345947265625, "t": 240.01333618164062 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 7, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 6, "text": "LM" }, { "bbox": { "b": 216.71160888671875, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 381.61065673828125, "t": 226.46434020996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 7, "text": "Character" }, { "bbox": { "b": 216.71160888671875, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 485.9234924316406, "t": 226.46434020996094 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 8, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 7, "text": "Reformer" }, { "bbox": { "b": 203.16160583496094, "coord_origin": "BOTTOMLEFT", "l": 339.2070007324219, "r": 373.1561279296875, "t": 212.91433715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 1, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 0, "start_row_offset_idx": 8, "text": "Custom" }, { "bbox": { "b": 203.16160583496094, "coord_origin": "BOTTOMLEFT", "l": 444.11981201171875, "r": 489.5780334472656, "t": 212.91433715820312 }, "col_span": 1, "column_header": false, "end_col_offset_idx": 2, "end_row_offset_idx": 9, "row_header": false, "row_section": false, "row_span": 1, "start_col_offset_idx": 1, "start_row_offset_idx": 8, "text": "Bio-Chem" } ] }, "footnotes": [], "label": "table", "parent": { "$ref": "#/body" }, "prov": [ { "bbox": { "b": 198.81939697265625, "coord_origin": "BOTTOMLEFT", "l": 332.6067810058594, "r": 499.9085388183594, "t": 338.24615478515625 }, "charspan": [ 0, 0 ], "page_no": 3 } ], "references": [], "self_ref": "#/tables/2" } ]

[]

{ "1": { "page_no": 1, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "2": { "page_no": 2, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "3": { "page_no": 3, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "4": { "page_no": 4, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "5": { "page_no": 5, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "6": { "page_no": 6, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "7": { "page_no": 7, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "8": { "page_no": 8, "size": { "height": 841.8900146484375, "width": 595.2760009765625 } }, "9": null, "10": null, "11": null, "12": null, "13": null, "14": null, "15": null, "16": null, "17": null, "18": null, "19": null, "20": null, "21": null, "22": null, "23": null, "24": null, "25": null, "26": null, "27": null, "28": null, "29": null, "30": null, "31": null, "32": null, "33": null, "34": null, "35": null, "36": null }