Upload .\data\raw\wiktionary_wiktionary_entries.jsonl with huggingface_hub

.//data//raw//wiktionary_wiktionary_entries.jsonl

@@ -0,0 +1,58 @@
+{"word": "!Xóõ", "text": "Icyongereza [ Hindura ] Wikipedia y'Icyongereza !Xóõ From the English Wiktionary: parts of this page need translation from English. Proper noun [ Hindura ] !Xóõ (en) a Khoisan language, known for its large phonemic inventory including complex click consonants. See also [ Hindura ] !Kung External links [ Hindura ] Inyandikorugero:ethnologue", "source": "wiktionary", "length": 341}
+{"word": "12", "text": "Izina [ Hindura ] Iyi nyandiko isobanura Cumi na kabiri mu mibare .", "source": "wiktionary", "length": 67}
+{"word": "Abanyarwanda", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Abaturage batuye u Rwanda . Iyo ari umwe bavuga Umunyarwanda", "source": "wiktionary", "length": 103}
+{"word": "Akanwa", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Igice cy'umubiri kiri ku mutwe . Akanwa niko abantu bakoresha bavuga , banywa , barya , bahumeka , ..", "source": "wiktionary", "length": 144}
+{"word": "Akarere k'ibiyaga bigari", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Akarere ko muri Afurika yo hagati kagizwe n'ibihugu bikikijwe n'ibiyaga binini cyangwa bigari. Ibyo bihugu birimo u Rwanda , Kongo , u Burundi , u Bugande , Tanzaniya Muri ibyo biyaga harimo Ikivu (cyangwa Ikiyaga cya Kivu ), Tanganyika , Vigitoriya , ...", "source": "wiktionary", "length": 298}
+{"word": "Amasano", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Reba isano .", "source": "wiktionary", "length": 55}
+{"word": "Amata", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Igitembabuzi kiva mu binyabuzimacyane cyane amata aba ari ay' Inka gishobora kunyobwa", "source": "wiktionary", "length": 128}
+{"word": "Amata y inka", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Ikinyobwa gikomoka ku Gukama Inka .", "source": "wiktionary", "length": 78}
+{"word": "Wikipedia", "text": "Icyongereza [ Hindura ] Izina [ Hindura ] Wikipedia (en) Wikipediya", "source": "wiktionary", "length": 67}
+{"word": "Wikipédia", "text": "Igifaransa [ Hindura ] Prononciation [ Hindura ] IPA : /wi.ki.pe.dja/ Izina [ Hindura ] Wikipédia (fr) Wikipediya", "source": "wiktionary", "length": 113}
+{"word": "Wiktionary", "text": "Icyongereza [ Hindura ] Izina [ Hindura ] Umushinga ufatanije uyobowe na Wikimedia Foundation kugirango utange inkoranyamagambo yubuntu kandi yuzuye mu ndimi zose; inkoranyamagambo, muri rusange, zakozwe nuwo mushinga.", "source": "wiktionary", "length": 218}
+{"word": "Zayire", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Zayire Reba Repubulika Iharanira Demukarasi ya Kongo", "source": "wiktionary", "length": 95}
+{"word": "abasore", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] abasore ... Nshakira abasore batatu bamfashe guterura uyu mufuka kuko uraremereye .", "source": "wiktionary", "length": 126}
+{"word": "abrigo", "text": "Igisipanyoli [ Hindura ] Izina [ Hindura ] abrigo m (es) ikote shelter", "source": "wiktionary", "length": 70}
+{"word": "ady", "text": "Malagasy [ Hindura ] intambara", "source": "wiktionary", "length": 30}
+{"word": "agua", "text": "Igisipanyoli [ Hindura ] Izina [ Hindura ] agua m (es) amazi", "source": "wiktionary", "length": 60}
+{"word": "akanwa", "text": "Redirect to: Akanwa", "source": "wiktionary", "length": 19}
+{"word": "akondro", "text": "Malagasy [ Hindura ] neke", "source": "wiktionary", "length": 25}
+{"word": "impuzamvugo", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Amagambo avugwa kimwe ariko atandikwa kimwe. Urugero: Iki --- Icyi avugwa kimwe ariko Iki rikoreshwa mu kwerekana ikintu kiri aho hafi naho Icyi ni igihe cy'umwaka gishyushye aho izuba riba riva cyane ndetse nta mvura igwa .", "source": "wiktionary", "length": 267}
+{"word": "imvura", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] imvura Amazi ava mu kirere akagwa ku butaka (ku isi ) mu buryo busanzwe cyangwa bwa cyimeza .", "source": "wiktionary", "length": 136}
+{"word": "imyaka", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Iri jambo rishobora gusobanura byinshi birimo: Ibihingwa mu murima Ubwinshi bw' umwaka : igipimo cy' igihe", "source": "wiktionary", "length": 149}
+{"word": "imyandikirwe", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Uko ijambo iri n'iri ryandikwa.", "source": "wiktionary", "length": 74}
+{"word": "informaticien", "text": "Igifaransa [ Hindura ] Prononciation [ Hindura ] IPA : /ɛ̃.fɔʁ.ma.ti.sjɛ̃/ SAMPA : /E~.fOR.ma.ti.sjE~/ Izina [ Hindura ] informaticien ( informaticiens , m ) (fr) mudasobwa specialist", "source": "wiktionary", "length": 183}
+{"word": "inkomamashyi", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Iri ijambo rikomoka ku gukoma amashyi . Inkomamashyi ni umuntu wemera agakora ibyo abwiwe byose atabanje kubitekerezaho. Impamvu zishobora gutera umuntu kuba inkomamashyi zirimo: ubwoba : umutegeka gukora ikintu aba ashobora kumugirira nabi (nko kumwica) hanyuma agahitamo kwemera agakora ibyo abwiwe byose ngo atavaho agirirwa nabi guhakirizwa : uyu ukoma amashyi ashobora kuba akunda guhakirizwa ngo arebe ko yahabwa ibyo yifuza kabone n'iyo yaba atabigenewe cyangwa atabikwiye. Ibi bikunze kugaragara nko muri politiki aho umuntu wifuza umwanya ahinduka inkomamashyi ku mutegetsi uyu n'uyu inda nini :", "source": "wiktionary", "length": 647}
+{"word": "intambara", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Ni intambara cyangwa imirwano.", "source": "wiktionary", "length": 73}
+{"word": "intangiriro", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Igihe kibanziriza ibindi mu kintu iki n'iki . Ingero [ Hindura ] Intangiriro y'umwaka ni ukwezi kwa Mutarama Intangiriro yo kubaho kwa buri muntu ni isama rya nyina Amasaku y'ibanze Ubutinde n'Amasaku [ Hindura ] intaângiriiro Gusemura mu ndimi [ Hindura ] Igifaransa: début (fr) , commencement (fr) Icyongereza: start (en) , beginning (en)", "source": "wiktionary", "length": 383}
+{"word": "intwaro", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Intwaro zikoreshwa mu ntambara.", "source": "wiktionary", "length": 74}
+{"word": "inyamaswa mbi ni umuntu", "text": "Ikinyarwanda [ Hindura ] Izina [ Hindura ] Uyu ni umugani w' umugenurano ushaka kwerekana ko umuntu ashobora kwitwara kinyamaswa .", "source": "wiktionary", "length": 130}
+{"word": "paix", "text": "Igifaransa [ Hindura ] Prononciation [ Hindura ] IPA : /pɛ/ SAMPA : /pE/ Izina [ Hindura ] paix (fr) amahoro", "source": "wiktionary", "length": 108}
+{"word": "papillon", "text": "Igifaransa [ Hindura ] Étymologie [ Hindura ] From Latin papilion Prononciation [ Hindura ] IPA : /pa.pi.jɔ̃/ Izina [ Hindura ] papillon (fr) ikinyugunyugu", "source": "wiktionary", "length": 155}
+{"word": "paraguas", "text": "Igisipanyoli [ Hindura ] Étymologie [ Hindura ] Izina [ Hindura ] paraguas (es) umutaka See also [ Hindura ] parabrisas paracaídas parasol , quitasol , sombrilla", "source": "wiktionary", "length": 161}
+{"word": "parapluie", "text": "Igifaransa [ Hindura ] Étymologie [ Hindura ] para + pluie ( rain ) Izina [ Hindura ] parapluie m (fr) umutaka (for rain)", "source": "wiktionary", "length": 121}
+{"word": "paz", "text": "Igiporutigali [ Hindura ] Izina [ Hindura ] paz f (pt) amahoro . Related terms [ Hindura ] pacificar pacífico pacifismo Igisipanyoli [ Hindura ] Étymologie [ Hindura ] Latin pax , pacis Prononciation [ Hindura ] IPA: /paθ/ Izina [ Hindura ] paz (es) amahoro Synonyms [ Hindura ] reposo Related terms [ Hindura ] pacífico pacifismo pacifista apaciguar", "source": "wiktionary", "length": 350}
+{"word": "pañuelo", "text": "Igisipanyoli [ Hindura ] Étymologie [ Hindura ] From paño + -uelo Izina [ Hindura ] pañuelo m (es) umushwara", "source": "wiktionary", "length": 108}
+{"word": "pollito", "text": "Igisipanyoli [ Hindura ] Étymologie [ Hindura ] From pollo + -ito Izina [ Hindura ] pollito (es) young inkoko young child", "source": "wiktionary", "length": 121}
+{"word": "pollo", "text": "Igitaliyani [ Hindura ] Izina [ Hindura ] pollo m ( bwinshi polli ) (it) inkoko (especially inkoko meat) Derived terms [ Hindura ] pollo delle pampas pollo sultano Igisipanyoli [ Hindura ] Étymologie [ Hindura ] Latin pullus Izina [ Hindura ] pollo (es) inkoko A Mexican paying to be smuggled illegally into the United States of America. Related terms [ Hindura ] polla f pollería f pollito m pollito feliz m See also [ Hindura ] coyote m gallina f gallo m", "source": "wiktionary", "length": 456}
+{"word": "poulet", "text": "Igifaransa [ Hindura ] Prononciation [ Hindura ] IPA : /pu.lɛ/ SAMPA : /pu.lE/ Izina [ Hindura ] poulet m (fr) inkoko (as a dish) See Also [ Hindura ]", "source": "wiktionary", "length": 150}
+{"word": "praofesora", "text": "Malagasy [ Hindura ] alimu", "source": "wiktionary", "length": 26}
+{"word": "курица", "text": "Ikirusiya [ Hindura ] Prononciation [ Hindura ] IPA : /ˈkurʲɪʦə/ Izina [ Hindura ] курица ( kúritsa , f ) (ru) inkoko (a bird of species Gallus gallus ). ( Derisive proverb ) Курица — не птица , баба — не человек . — A chicken is not a bird and a woman is not a human. ( Same pattern ) Курица — не птица, Польша — не заграница . — A chicken is not a bird and Poland is not a foreign country. (Poland here may be replaced by any other formerly socialist country or former Soviet republic,) hen (female chicken) . курица клохчет ( кудахчет ) — a hen cackles ( clucks ) куры несутся — hens lay eggs курица снесла яйцо — a hen has laid an egg курица высиживает яйца — а hen is brooding мокрая курица — a weak and pitiful -looking person ( literally ‘wet chicken’) . Declension [ Hindura ] singular plural 1 plural 2 nominative ку́рица ку́ры ку́рицы genitive ку́рицы кур ку́риц dative ку́рице ку́рам ку́рицам accusative ку́рицу кур ку́риц instrumental ку́рицей ку́рами ку́рицами prepositional о ку́рице о ку́рах о ку́рицах Related terms [ Hindura ] Nouns [ Hindura ] кур курёнок курочка Adjectives [ Hindura ] курий куриный курицын курячий Derived terms [ Hindura ] денег куры не клюют как курица лапой как курица с яйцом носится курам на смех See also [ Hindura ] клуша наседка петух цыплёнок", "source": "wiktionary", "length": 1288}
+{"word": "масло", "text": "Ikirusiya [ Hindura ] Izina [ Hindura ] масло ( máslo , n , масла , maslá , pl ) (ru) amavuta oil (vegetable or mineral) Declension [ Hindura ] singular plural nominative ма́сло масла́ genitive ма́сла ма́сел dative ма́слу масла́м accusative ма́сло масла́ instrumental ма́слом масла́ми prepositional о ма́сле о масла́х Related terms [ Hindura ] масляный See also [ Hindura ] нефть", "source": "wiktionary", "length": 379}
+{"word": "мир", "text": "Ikirusiya [ Hindura ] Izina [ Hindura ] мир ( mir , m , миры , mirý , pl ) (ru) amahoro война́ и мир — war and peace мир во всём мире (mir vo vsjom míre) — world peace universe происхождение мира (proiskhoždénije míra) — origin of the universe planet Declension [ Hindura ] singular plural nominative мир миры́ genitive ми́ра миро́в dative ми́ру мира́м accusative мир миры́ instrumental ми́ром мира́ми prepositional о ми́ре о мира́х locative в миру́ в мира́х", "source": "wiktionary", "length": 458}
+{"word": "мотылёк", "text": "Ikirusiya [ Hindura ] Izina [ Hindura ] мотылёк ( motyljók , m , мотыльки , motyl’kí , pl ) (ru) moth (the small variety that swarms at night) (colloquial) ikinyugunyugu Declension [ Hindura ] singular plural nominative мотылёк мотыльки́ genitive мотылька́ мотылько́в dative мотыльку́ мотылька́м accusative мотылька́ мотылько́в instrumental мотылько́м мотылька́ми prepositional о мотыльке́ о мотылька́х Synonyms [ Hindura ] бабочка моль", "source": "wiktionary", "length": 436}
+{"word": "мука", "text": "Ikirusiya [ Hindura ] Izina [ Hindura ] мука ( múka , f , муки , múki , pl ) (ru) pain , anguish , torment , suffering , torture harassment Declension [ Hindura ] singular plural nominative му́ка му́ки genitive му́ки мук dative му́ке му́кам accusative му́ку му́ки instrumental му́кой, -ою му́ками prepositional о му́ке о му́ках Izina [ Hindura ] мука ( muká , f , mass noun , no plural ) (ru) ifu , meal Declension [ Hindura ] singular nominative мука́ genitive муки́ dative муке́ accusative муку́ instrumental муко́й, -о́ю prepositional о муке́", "source": "wiktionary", "length": 545}
+{"word": "мёд", "text": "Ikirusiya [ Hindura ] Izina [ Hindura ] мёд ( mjod , m ) (ru) ubuki mead Usage notes [ Hindura ] Plural of мёд is used when referring to the sorts of honey (or mead). Declension [ Hindura ] Diminutive forms [ Hindura ] медок Derived terms [ Hindura ] Adjectives [ Hindura ] медовый медоносный медоточивый Nouns [ Hindura ] медведь медовар медоварение медовуха медовый месяц медогонка медонос медосбор", "source": "wiktionary", "length": 400}
+{"word": "пальто", "text": "Ikirusiya [ Hindura ] Izina [ Hindura ] пальто ( pal’tó , n , indeclinable ) (ru) ikote", "source": "wiktionary", "length": 87}
+{"word": "существительное", "text": "Ikirusiya [ Hindura ] Prononciation [ Hindura ] IPA : /suʃʲʃʲɪˈstvʲitʲɪlʲnəjə/ Izina [ Hindura ] существительное ( suščestvítel’noje , n ) (ru) izina , substantive Synonyms [ Hindura ] имя существительное Related terms [ Hindura ] существительный Declension [ Hindura ] Singular Plural Nominative существи́тельное существи́тельные genitive существи́тельного существи́тельных dative существи́тельному существи́тельным accusative существи́тельное существи́тельные instrumental существи́тельным существи́тельными prepositional о существи́тельном о существи́тельных", "source": "wiktionary", "length": 561}
+{"word": "اسم", "text": "Icyarabu [ Hindura ] Izina [ Hindura ] اسم ( ’ism , m , اسماء , ’asmā’ , pl , اسام , ’asāmin ) (ar) name بسم الله (b-ism illāh) — in the name of God بسم الله الرحمن الرحيم (b-ism-illāh ir-raħmān ir-raħīm) — in the name of God, the Merciful, the Compassionate izina appellation reputation , prestige Related terms [ Hindura ] اسمي Farsi [ Hindura ] Izina [ Hindura ] === اسم ( esm ) (fa) name izina Urdu [ Hindura ] Izina [ Hindura ] اسم (esam) name substantive Synonym [ Hindura ] نام الحوطي", "source": "wiktionary", "length": 491}
+{"word": "اٹا", "text": "Urdu [ Hindura ] Izina [ Hindura ] اٹا ( āt'ā , m ) (ur) ifu meal dust Devanagari spelling [ Hindura ] आटा", "source": "wiktionary", "length": 106}
+{"word": "버터", "text": "Igikoreya [ Hindura ] Étymologie [ Hindura ] From English. Izina [ Hindura ] 버터 ( beoteo ) (ko) amavuta", "source": "wiktionary", "length": 103}
+{"word": "빠르다", "text": "Igikoreya [ Hindura ] Étymologie [ Hindura ] From English. Izina [ Hindura ] 빠르다 ( ppareuda ) (ko) vuba", "source": "wiktionary", "length": 103}
+{"word": "소", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 소 ( so ) (ko) inka (substance)", "source": "wiktionary", "length": 70}
+{"word": "아니오", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 아니오 ( anio ) (ko) oya", "source": "wiktionary", "length": 61}
+{"word": "아버지", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 아버지 (' abeoji' ) (ko) so", "source": "wiktionary", "length": 64}
+{"word": "안녕", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 안녕 ( annyeong ) (ko) amahoro , tranquility", "source": "wiktionary", "length": 82}
+{"word": "알다", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 알다 ( alda ) (ko) nabizi", "source": "wiktionary", "length": 63}
+{"word": "여자", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 여자 (' yeoja' ) (ko) umugore", "source": "wiktionary", "length": 67}
+{"word": "영어", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 영어 ( 英語 , yeongeo ) (ko) Icyongereza", "source": "wiktionary", "length": 76}
+{"word": "우산", "text": "Igikoreya [ Hindura ] Izina [ Hindura ] 우산 ( 雨傘 , usan ) (ko) umutaka", "source": "wiktionary", "length": 69}