{"id":"https://openalex.org/W7225377301","doi":"https://doi.org/10.5281/zenodo.24023073","title":"Modern Hebrew document line recognition (printed and handwriting-style), trained from scratch on synthetic pages; recognizer of the HeOCR R4d + A1 + R3h pipeline","display_name":"Modern Hebrew document line recognition (printed and handwriting-style), trained from scratch on synthetic pages; recognizer of the HeOCR R4d + A1 + R3h pipeline","publication_year":2026,"publication_date":"2026-11-07","ids":{"openalex":"https://openalex.org/W7225377301","doi":"https://doi.org/10.5281/zenodo.24023073"},"language":"en","primary_location":{"id":"doi:10.5281/zenodo.24023073","doi":"https://doi.org/10.5281/zenodo.24023073","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24023073","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"volume":null,"type":"article","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.5281/zenodo.24023073","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5161045255","display_name":"Shay Palachy Affek","orcid":"https://orcid.org/0000-0002-1459-9320","observed_orcids":["https://orcid.org/0000-0002-1459-9320"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Palachy Affek, Shay","raw_affiliation_strings":[],"raw_orcid":"https://orcid.org/0000-0002-1459-9320","affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5161072239","display_name":"HeOCR maintainers","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"HeOCR maintainers","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.0,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10601","display_name":"Handwritten Text Recognition Techniques","score":1.0,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10601","display_name":"Handwritten Text Recognition Techniques","score":1.0,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12707","display_name":"Vehicle License Plate Recognition","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/modern-hebrew","display_name":"Modern Hebrew","score":0.4359999895095825},{"id":"https://openalex.org/keywords/handwriting-recognition","display_name":"handwriting recognition","score":0.6399999856948853},{"id":"https://openalex.org/keywords/printed-text-recognition","display_name":"printed character recognition","score":0.9290000200271606},{"id":"https://openalex.org/keywords/synthetic-training-data","display_name":"synthetic training data","score":0.746999979019165},{"id":"https://openalex.org/keywords/kraken","display_name":"Kraken","score":0.503000020980835},{"id":"https://openalex.org/keywords/line-segmentation","display_name":"line segmentation","score":0.5419999957084656},{"id":"https://openalex.org/keywords/reading-order","display_name":"reading order","score":0.8399999737739563},{"id":"https://openalex.org/keywords/digits","display_name":"digits","score":0.46799999475479126}],"concepts":[{"id":"https://openalex.org/C2781235140","wikidata":"https://www.wikidata.org/wiki/Q275131","display_name":"Scratch","level":2,"score":0.6866000294685364},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.675599992275238},{"id":"https://openalex.org/C43521106","wikidata":"https://www.wikidata.org/wiki/Q2165493","display_name":"Pipeline (software)","level":2,"score":0.5845000147819519},{"id":"https://openalex.org/C43364308","wikidata":"https://www.wikidata.org/wiki/Q8799","display_name":"Byte","level":2,"score":0.5138000249862671},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.45579999685287476},{"id":"https://openalex.org/C198352243","wikidata":"https://www.wikidata.org/wiki/Q37105","display_name":"Line (geometry)","level":2,"score":0.43549999594688416},{"id":"https://openalex.org/C554936623","wikidata":"https://www.wikidata.org/wiki/Q199657","display_name":"Reading (process)","level":2,"score":0.4246000051498413},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.4198000133037567},{"id":"https://openalex.org/C91304198","wikidata":"https://www.wikidata.org/wiki/Q9288","display_name":"Hebrew","level":2,"score":0.4065000116825104},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.364300012588501},{"id":"https://openalex.org/C183469790","wikidata":"https://www.wikidata.org/wiki/Q333501","display_name":"Crash","level":2,"score":0.3564999997615814},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.33959999680519104},{"id":"https://openalex.org/C2778802812","wikidata":"https://www.wikidata.org/wiki/Q79894","display_name":"Braille","level":2,"score":0.2827000021934509},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.27649998664855957},{"id":"https://openalex.org/C519991488","wikidata":"https://www.wikidata.org/wiki/Q28865","display_name":"Python (programming language)","level":2,"score":0.2687999904155731},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.26510000228881836},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.263700008392334},{"id":"https://openalex.org/C97250363","wikidata":"https://www.wikidata.org/wiki/Q235557","display_name":"File format","level":2,"score":0.2515000104904175}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.5281/zenodo.24023073","doi":"https://doi.org/10.5281/zenodo.24023073","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24023073","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},{"id":"doi:10.5281/zenodo.24023074","doi":"https://doi.org/10.5281/zenodo.24023074","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24023074","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"doi:10.5281/zenodo.24023073","doi":"https://doi.org/10.5281/zenodo.24023073","is_oa":true,"landing_page_url":"https://doi.org/10.5281/zenodo.24023073","pdf_url":null,"source":{"id":"https://openalex.org/S4306400562","display_name":"Zenodo (CERN European Organization for Nuclear Research)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I67311998","host_organization_name":"European Organization for Nuclear Research","host_organization_lineage":["https://openalex.org/I67311998"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"article"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7225423610","https://openalex.org/W7225417216","https://openalex.org/W7172376695","https://openalex.org/W7172432031","https://openalex.org/W385765767","https://openalex.org/W3208193076","https://openalex.org/W7172396795","https://openalex.org/W7223032124","https://openalex.org/W7047328609","https://openalex.org/W7204490102"],"abstract_inverted_index":{"HeOCR":[0,33,139,419],"R3h:":[1],"Hebrew":[2,18,310,382,411,448],"document":[3,798],"line":[4,158,168,205,228,415,633,683,699,712,941,965,984,993,1301,1304],"recognition":[5,12,45,120,263,379],"(kraken)":[6],"R3h":[7,42,65,258,707,1179,1220],"is":[8,43,85,97,369,398,452,480,517,573,584,620,716,836,849,913,942,988,1154,1159,1297,1313,1390,1397],"a":[9,86,90,226,236,290,496,592,621,697,745,977,1141,1289,1375,1398],"kraken":[10,124,130,187,216,244,503,589,602,641,662,687,703,722,762,771,1230],"text":[11,223,483,511,530,625,653,693,801,887,925],"model":[13,182,298,563,1337],"for":[14,194,380,410,648,658,719,896,1281],"lines":[15,55,286,742,1285,1295],"of":[16,47,94,292,422,426,594,946,979,995,1170,1196,1256,1275,1334,1362,1382,1387,1395],"modern":[17,381],"documents.":[19,428],"It":[20,84,397,848,1043],"reads":[21,66,741,1272],"printed":[22,827,832],"and":[23,30,127,143,318,325,333,374,390,417,434,476,524,535,543,546,675,759,795,829,842,859,878,902,920,923,930,932,1000,1019,1075,1085,1129,1267,1270,1279,1345,1358],"handwriting-style":[24,395],"lines,":[25,1269],"including":[26],"digits,":[27],"Latin":[28,327,547,927],"letters":[29,328],"typographic":[31,353,527,936,1198],"punctuation.":[32,937],"trained":[34,126,212,1111],"it":[35,489,680,1265,1271,1296],"from":[36,170,235,540,807,1003,1093,1100],"scratch":[37],"on":[38,98,104,213,439,643,775,1057,1070,1149,1294,1370,1379,1391],"synthetic":[39,99,440,1348],"pages":[40,441,806,839,845,1169,1352],"only.":[41,442],"the":[44,54,203,218,232,308,322,326,364,402,435,455,515,529,554,562,574,576,627,655,681,689,905,957,962,969,992,1048,1063,1076,1101,1105,1132,1168,1204,1260,1291,1354,1359],"step":[46],"HeOCR's":[48,1282],"three-model":[49],"pipeline:":[50,1174,1290],"-":[51,57,64,197,221,307,321,331,345,352,363,669,691],"R4d":[52,403,705,1175,1278],"finds":[53],"(segmentation);":[56],"A1":[58,184,191,210,405,569,1177,1280],"puts":[59],"them":[60,1273],"in":[61,295,392,401,454,484,518,629,709,870,944,1320],"reading":[62,200,1145],"order;":[63],"them.":[67,532],"The":[68,82,151,297,564,652,834,873,911,986,1157,1173,1249,1317,1347,1350],"three":[69,947],"are":[70,287,438,469,526,538,672,869,885,1078,1097,1252],"released":[71,1181],"together,":[72],"as":[73,112,1364],"separate":[74],"records":[75],"that":[76,468,659,720,743,1096,1383,1396],"link":[77],"to":[78,207,289,490,508,558,638,748,904,981,990],"each":[79,171,1171],"other":[80,446],"(see":[81,141,769],"pipeline).":[83],"research":[87,408],"baseline,":[88],"not":[89,453,464,470,1079,1259,1288,1327,1401],"production":[91],"transcriber.":[92],"All":[93,790],"its":[95,198,222,764,782,973,1058],"evaluation":[96,103,432,1094,1351],"pages;":[100],"no":[101,370,375,760,796,899],"reviewed":[102,430],"real":[105,427,797,1371],"scans":[106],"exists":[107],"yet.":[108],"File":[109],"heocr-r3h-recognition.safetensors":[110,254,615,735],"(trained":[111],"best_0.9959.safetensors),":[113],"16,167,536":[114],"bytes":[115,1077],"sha256":[116,882],"f76affa84e29536aa1d41f89d3507183c104f1620ca953fdfa5dd4b7a00ba6be":[117],"Model":[118,154],"type":[119,174],"(line-level,":[121],"CTC)":[122],"Format":[123],"safetensors;":[125],"tested":[128],"with":[129,315,495,504,561,640,898,926,935,953,961,968,1037,1062,1104],"/":[131,347],"ketos":[132,260,1011],"7.0.2":[133,188,642,665,772,1012],"Licence":[134],"Apache-2.0,":[135],"Copyright":[136],"(c)":[137],"2026":[138],"maintainers":[140],"LICENSE":[142],"NOTICE)":[144],"Training":[145,487,786,788,998],"record":[146],"r3h_training_run.json":[147],"(heocr_engine_training_run.v1)":[148],"Code":[149],"HeOCR/heocr-engine":[150],"pipeline":[152,234,566,678],"Step":[153],"Record":[155],"Segmentation:":[156],"baselines,":[157],"polygons,":[159],"typed":[160],"regions":[161],"R4d,":[162],"heocr-r4d-segmentation.safetensors":[163,248,607,731],"(d32cd1a8\u2026)":[164],"10.5281/zenodo.24023070":[165],"Reading":[166],"order:":[167],"order":[169,201,229,552,583,700,1146],"line's":[172],"region":[173,214,252,612,673],"A1,":[175,208],"heocr-a1-reading-order.safetensors":[176,250,610],"(69aa1946\u2026)":[177],"10.5281/zenodo.24023072":[178],"Recognition":[179],"R3h,":[180],"this":[181],"10.5281/zenodo.24023074":[183],"needs":[185],"heocr-engine.":[186],"cannot":[189,666],"run":[190,502,596],"by":[192,586,853,889,950,976,1340,1374],"itself,":[193],"two":[195,1107,1356],"reasons:":[196],"neural":[199],"passes":[202],"segmenter's":[204,682],"types":[206,684],"while":[209],"was":[211,793,802,1050,1110,1147],"types.":[215],"rejects":[217,688],"class-mapping":[219],"mismatch;":[220],"output":[224,516,694],"ignores":[225,696],"learned":[227,698],"(mittagessen/kraken#820).":[230,701],"Run":[231],"whole":[233],"heocr-engine":[237,243,601,1014],"checkout":[238,593],"(pip":[239],"install":[240,598],"'.[kraken]'):":[241],"sh":[242],"--api":[245,603],"python":[246,604],"--segmentation-model":[247,606],"--reading-order-model":[249,609],"--reading-order-classes":[251,611],"--recognition-model":[253,614],"page.jpg":[255,617,724],"page.json":[256,618,619],"Architecture":[257],"uses":[259,763],"7.0.2's":[261],"default":[262,765,1216],"network,":[264],"unchanged:":[265],"[1,120,0,1":[266],"Cr3,13,32":[267,270],"Do0.1,2":[268,271,274,277,280,282],"Mp2,2":[269,272,275],"Cr3,9,64":[273,276],"S1(1x0)1,3":[278],"Lbx200":[279,281,283],"Do]":[284],"Input":[285],"scaled":[288],"height":[291,974],"120":[293],"pixels,":[294],"greyscale.":[296],"expects":[299],"baseline":[300,409],"segmentation":[301],"(seg_type:":[302],"baselines).":[303],"Codec":[304],"(117":[305],"symbols):":[306],"27":[309],"letters,":[311,384,915],"final":[312],"forms":[313,389,919],"included,":[314],"geresh":[316,536],"\u05f3":[317,537],"gershayim":[319],"\u05f4;":[320],"digits":[323,929],"0\u20139":[324],"A\u2013Z":[329],"a\u2013z;":[330],"space":[332],"ASCII":[334,541],"punctuation:":[335,354],"!":[336],"\"":[337,359,360,361,542],"%":[338],"&":[339],"'":[340,357,358],"(":[341],")":[342],"+":[343,404,568,570,706,1176,1178,1219],",":[344],".":[346],":":[348],";":[349],"?":[350],"@;":[351],"\u2013":[355],"\u2014":[356],"\u2026;":[362],"shekel":[365],"sign":[366],"\u20aa.":[367],"There":[368],"niqqud":[371],"(vowel":[372],"points)":[373],"cantillation.":[376],"Uses":[377],"Line":[378,938],"documents:":[383],"notes,":[385,916],"archive":[386,816,819],"cards,":[387],"ledgers,":[388,918,1130],"notices,":[391],"print":[393],"or":[394,445,799,1112],"fonts.":[396,1108],"best":[399,1049],"used":[400,531],"pipeline.":[406],"A":[407,1330],"OCR:":[412],"comparing":[413],"segmenters,":[414],"geometries":[416],"later":[418],"models.":[420],"Out":[421],"scope":[423],"Unsupervised":[424],"transcription":[425],"No":[429,714,1005],"real-page":[431],"exists,":[433],"scores":[436,1367],"below":[437,579],"Medieval,":[443],"liturgical":[444],"historical":[447],"scripts.":[449],"Niqqud,":[450],"which":[451],"codec.":[456],"Real":[457],"handwriting.":[458],"\"Handwriting-style\"":[459],"means":[460],"one":[461,632,649,945,1314],"handwriting":[462],"font,":[463],"human":[465],"hands.":[466],"Documents":[467],"mainly":[471],"Hebrew.":[472],"Transcription":[473],"guidelines,":[474],"normalization":[475],"transformations":[477],"Ground":[478],"truth":[479],"Unicode":[481],"NFC":[482],"logical":[485,630],"order.":[486,520,713,1276],"reorders":[488],"display":[491,519],"order,":[492,631],"kraken's":[493,676,692,710,954,1215],"convention,":[494,1400],"right-to-left":[497],"base":[498],"direction.":[499],"At":[500],"inference,":[501],"--reorder":[505,736,750,1227],"--base-dir":[506,737,751,1228],"R":[507,738,752],"get":[509,559],"logical-order":[510,555,754],"back.":[512],"Without":[513],"it,":[514],"Punctuation:":[521],"quotes,":[522],"dashes":[523],"ellipses":[525],"where":[528],"Gershayim":[533],"\u05f4":[534],"distinct":[539],"'.":[544],"Digits":[545],"words":[548],"keep":[549],"their":[550],"left-to-right":[551],"inside":[553],"line.":[556],"How":[557],"started":[560],"full":[565],"(R4d":[567],"R3h).":[571],"This":[572,1312],"configuration":[575],"headline":[577,1183],"figures":[578],"come":[580],"from.":[581],"A1's":[582,670,1144],"applied":[585],"heocr-engine's":[587,1321],"in-process":[588],"stage.":[590],"In":[591],"heocr-engine,":[595],"pip":[597],"'.[kraken]',":[599],"then:":[600],"\\":[605,608,613,616,732],"prediction":[622,657],"record.":[623],"Its":[624],"holds":[626],"page":[628,660,852],"per":[634,851],"row.":[635],"Tested":[636],"end":[637,639],"CPU":[644],"(about":[645],"8":[646,1046],"s":[647],"seed-101":[650,1115,1187,1234,1306],"page).":[651],"matched":[654],"scored":[656],"exactly.":[661],"alone.":[663],"Kraken":[664],"use":[667,1353],"A1:":[668],"classes":[671],"types,":[674],"own":[677,1059],"gives":[679],"instead,":[685],"so":[686],"pair.":[690],"also":[695],"With":[702,756],"alone,":[704,1293],"runs":[708],"heuristic":[711],"figure":[715],"reported":[717],"here":[718],"configuration.":[721],"-i":[723,730],"page.txt":[725],"segment":[726,757],"-bl":[727,758,1222],"-d":[728,739,1223],"horizontal-rl":[729,740],"ocr":[733,1226],"-m":[734],"share":[744],"row":[746],"right":[747],"left.":[749],"returns":[753],"text.":[755],"-i,":[761],"blla":[766,1218,1250],"segmenter":[767],"instead":[768],"below).":[770],"sometimes":[773],"aborts":[774],"exit":[776],"(\"recursive_mutex":[777],"lock":[778],"failed\")":[779],"after":[780,1045],"writing":[781],"output.":[783],"Retrying":[784],"works.":[785],"details":[787],"data":[789,1088],"synthetic.":[791,1155],"Nothing":[792],"scraped,":[794],"third-party":[800,900,1006],"used.":[803],"Pages:":[804],"3,000":[805],"hocrsyngen":[808,874,1090],"at":[809],"88ba5e6,":[810],"seed":[811,1068,1333],"7201,":[812],"over":[813,1167],"seven":[814],"templates:":[815],"card,":[817,820],"faded":[818],"handwritten":[821,824],"note,":[822,825],"worn":[823],"ledger,":[826],"letter":[828],"heavily":[830],"scanned":[831],"letter.":[833],"split":[835,1061],"2,850":[837],"training":[838,1102,1315,1328,1332],"(47,633":[840],"lines)":[841],"150":[843],"validation":[844,1060],"(2,706":[846],"lines).":[847],"assigned":[850],"hash.":[854],"Fonts:":[855],"Alef":[856],"(2,143":[857],"pages)":[858],"Gveret":[860],"Levin":[861],"(857":[862],"pages),":[863],"both":[864],"SIL":[865],"OFL":[866],"1.1.":[867],"Credits":[868],"NOTICE.":[871],"Text:":[872],"corpora":[875,1095],"S10b,":[876],"S13a":[877],"S13b":[879],"(932,130":[880],"bytes;":[881],"7c2a62b8\u2026).":[883],"They":[884],"original":[886],"authored":[888],"AI":[890],"agents":[891],"(Claude":[892],"Opus":[893],"5.5":[894],"sessions)":[895],"HeOCR,":[897],"text,":[901,1103],"dedicated":[903],"public":[906],"domain":[907],"under":[908,1161],"CC0":[909],"1.0.":[910],"register":[912],"administrative":[914],"archives,":[917],"notices;":[921],"procurement":[922],"technical":[924],"terms,":[928],"codes;":[931],"typeset":[933],"prose":[934],"geometry:":[939],"Each":[940],"cropped":[943],"ways,":[948],"chosen":[949],"hash:":[951],"3/4":[952],"polygonizer":[955],"around":[956],"true":[958,1300],"baseline;":[959],"1/8":[960,967],"generator's":[963],"exact":[964],"polygon;":[966],"polygon's":[970],"bounding":[971],"box,":[972],"adjusted":[975],"pad":[978],"\u22120.14":[980],"approximate":[982],"Tesseract":[983,1303],"boxes.":[985],"mix":[987],"meant":[989],"cover":[991],"geometry":[994,1065],"different":[996],"segmenters.":[997],"procedure":[999],"hyperparameters":[1001],"Initialisation:":[1002],"scratch.":[1004],"weights":[1007],"were":[1008],"loaded.":[1009],"Trainer:":[1010],"through":[1013],"train-kraken":[1015],"(heocr-engine":[1016],"55b0f7d).":[1017],"Optimiser":[1018],"data:":[1020],"AdamW,":[1021],"learning":[1022],"rate":[1023],"0.001,":[1024],"constant":[1025],"schedule;":[1026],"batch":[1027],"size":[1028],"16,":[1029,1031],"padding":[1030],"augmentation":[1032],"on.":[1033,1114],"Stopping:":[1034],"early":[1035],"stopping":[1036],"lag":[1038],"3":[1039],"(minimum":[1040],"5":[1041],"epochs).":[1042],"stopped":[1044],"epochs;":[1047],"epoch":[1051],"4.":[1052],"Validation:":[1053],"character":[1054],"accuracy":[1055,1369],"0.9959,":[1056],"same":[1064,1106,1355,1360],"mix.":[1066],"Run:":[1067],"42,":[1069],"Apple":[1071],"mps.":[1072],"Not":[1073],"deterministic,":[1074],"reproducible.":[1080],"Evaluation":[1081],"Testing":[1082,1087],"data,":[1083],"factors":[1084],"metrics":[1086],"Batches:":[1089],"batches":[1091],"rendered":[1092],"kept":[1098],"apart":[1099],"None":[1109],"tuned":[1113,1148],"(the":[1116,1180],"long-running":[1117],"benchmark);":[1118],"held-out;":[1119],"eval-mixed-heldout":[1120,1191,1238,1339],"(mixed":[1121],"Hebrew/Latin/digits);":[1122],"eval-typo-heldout":[1123,1193,1240],"(typographic":[1124],"punctuation);":[1125],"eval-layout-heldout":[1126,1201,1242],"(archive":[1127],"cards":[1128],"plus":[1131],"four":[1133],"benchmark":[1134],"templates).":[1135],"hocrgen":[1136,1210,1244],"fixture:":[1137],"hocrgen's":[1138,1162],"fixture":[1139,1211,1245],"batch,":[1140],"development":[1142],"set.":[1143],"it.":[1150],"Gold:":[1151],"all":[1152],"gold":[1153,1206,1399],"Metrics":[1156],"metric":[1158],"CER":[1160,1186,1233],"primary.v1":[1163],"normalisation":[1164],"profile,":[1165],"pooled":[1166],"batch.":[1172],"models;":[1182],"figures):":[1184],"Batch":[1185,1232,1299],"0.020":[1188],"held-out":[1189,1236,1309],"0.010":[1190],"0.017":[1192],"0.009":[1194],"(269":[1195],"308":[1197],"marks":[1199],"kept)":[1200],"0.124":[1202],"(mostly":[1203],"ledger":[1205,1392],"convention;":[1207],"see":[1208],"below)":[1209],"(development":[1212,1246],"set)":[1213,1247],"0.130":[1214],"segmenter:":[1217],"(segment":[1221],"horizontal-rl,":[1224],"then":[1225],"R,":[1229],"7.0.2):":[1231],"0.168":[1235],"0.238":[1237],"0.230":[1239],"0.291":[1241],"0.498":[1243],"0.404":[1248],"rows":[1251],"much":[1253],"worse":[1254],"because":[1255],"blla's":[1257],"segmentation,":[1258],"recognizer.":[1261],"On":[1262],"these":[1263],"layouts":[1264],"splits":[1266],"merges":[1268],"out":[1274],"Use":[1277],"figures.":[1283],"Given":[1284],"(a":[1286],"diagnostic,":[1287],"recognizer":[1292],"handed):":[1298],"polygons":[1302],"boxes":[1305],"0.014":[1307],"0.116":[1308],"0.004":[1310],"0.088":[1311],"run.":[1316],"paired":[1318],"intervals":[1319],"docs/baselines.md":[1322],"(R3h":[1323],"section)":[1324],"resample":[1325],"pages,":[1326],"runs.":[1329],"second":[1331],"an":[1335],"earlier":[1336],"moved":[1338],"about":[1341],"0.005.":[1342],"Bias,":[1343],"risks":[1344],"limitations":[1346],"gap.":[1349],"fonts":[1357],"kinds":[1361],"degradation":[1363],"training.":[1365],"These":[1366],"overstate":[1368],"scans,":[1372],"possibly":[1373],"wide":[1376],"margin.":[1377],"Ledgers":[1378],"eval-layout-heldout:":[1380],"Most":[1381,1394],"batch's":[1384],"error,":[1385],"2,432":[1386],"2,611":[1388],"edits,":[1389],"pages.":[1393],"misreading.":[1402]},"counts_by_year":[],"updated_date":"2026-11-08T12:12:28.789561","created_date":"2026-11-07T00:00:00"}