Upload 7 files
Browse files- src/core/article_urls +399 -0
- src/core/clean_core.jsonl +0 -0
- src/core/core.jsonl +0 -0
- src/core/preprocessed_core.jsonl +0 -0
- src/core/united_core.txt +0 -0
- src/tokenizers_analysis.py +89 -0
- src/tokenizers_trainer.py +239 -0
src/core/article_urls
ADDED
|
@@ -0,0 +1,399 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
https://lenta.ru/news/2025/10/10/v-kremle-nazvali-chudovischnymi-zayavleniya-zelenskogo-o-raketah-tomahawk/
|
| 2 |
+
https://lenta.ru/news/2025/10/10/56-letnyaya-supermodel-v-prozrachnom-platie-bez-nizhnego-belya-posetila-balet/
|
| 3 |
+
https://lenta.ru/news/2025/10/10/musorovoz-pereehal-pensionerku-v-rossiyskom-gorode-i-popal-na-video/
|
| 4 |
+
https://lenta.ru/news/2025/10/10/zhenschina-nedelyu-probolela-doma-i-razbogatela-na-vosem-millionov-rubley/
|
| 5 |
+
https://lenta.ru/news/2025/10/10/iphone17-battery/
|
| 6 |
+
https://lenta.ru/news/2025/10/10/v-rossii-raskryli-tseli-udara-po-severu-ukrainy/
|
| 7 |
+
https://lenta.ru/news/2025/10/10/v-rossiyskom-gorode-poymali-napavshego-na-devushku-na-pustyre-manyaka/
|
| 8 |
+
https://lenta.ru/news/2025/10/10/byvshiy-sovladelets-banka-oktrytie-obvinen-v-hischenii-34-milliardov-rubley-iz-banka/
|
| 9 |
+
https://lenta.ru/news/2025/10/10/opublikovan-obnovlennyy-reyting-putina-sredi-rossiyan/
|
| 10 |
+
https://lenta.ru/news/2025/10/10/rabotodateli-stali-chasche-upominat-odin-navyk-pri-poiske-sotrudnikov/
|
| 11 |
+
https://lenta.ru/news/2025/10/10/stali-izvestny-podrobnosti-taynogo-vozvrascheniya-plennyh-rossiyskih-snayperov-iz-plena-vsu/
|
| 12 |
+
https://lenta.ru/news/2025/10/10/raskryta-detal/
|
| 13 |
+
https://lenta.ru/news/2025/10/10/zarubivshiy-toporom-pyateryh-chelovek-izhevskiy-raskolnikov-vyshel-na-svobodu/
|
| 14 |
+
https://lenta.ru/news/2025/10/10/passazhir-samoleta-razdelsya-i-popytalsya-iznasilovat-poputchitsu-vo-vremya-poleta/
|
| 15 |
+
https://lenta.ru/news/2025/10/10/rossiyanin-povstrechalsya-s-kitayankami-i-opisal-otnosheniya-s-nimi-slovami-eto-bylo-udushayusche/
|
| 16 |
+
https://lenta.ru/news/2025/10/10/putin-progulyalsya-v-kompanii-alieva-i-drugih-liderov-sng-i-popal-na-video/
|
| 17 |
+
https://lenta.ru/news/2025/10/10/killer-s-posobnikom-za-300-dollarov-raspravilis-s-direktorom-litseya/
|
| 18 |
+
https://lenta.ru/news/2025/10/10/glava-minzdrava-ssha-rasskazal-o-svyazi-mezhdu-obrezaniem-i-autizmom/
|
| 19 |
+
https://lenta.ru/news/2025/10/10/prezidenta-odnoy-strany-otstranili-ot-dolzhnosti/
|
| 20 |
+
https://lenta.ru/news/2025/10/10/alesya-kafelnikova-v-poluprozrachnoy-yubke-prokatilas-na-loshadi/
|
| 21 |
+
https://lenta.ru/news/2025/10/10/vrach-raskryla-dva-deystvennyh-sposoba-umenshit-podtekaniya-mochi/
|
| 22 |
+
https://lenta.ru/news/2025/10/10/malkin-vyshel-na-30-e-mesto-v-spiske-luchshih-bombardirov-v-istorii-nhl/
|
| 23 |
+
https://lenta.ru/news/2025/10/10/byvshaya-koroleva-krasoty-popalas-s-358-kilogrammami-narkotikov-i-izbezhala-tyurmy/
|
| 24 |
+
https://lenta.ru/news/2025/10/10/fedoseevu-shukshinu-ekstrenno-gospitalizirovali-v-moskve/
|
| 25 |
+
https://lenta.ru/news/2025/10/10/kaprizov-sdelal-assistentskiy-het-trik-v-matche-nhl/
|
| 26 |
+
https://lenta.ru/news/2025/10/10/izvestnyy-komik-perenes-srochnuyu-operatsiyu-i-nazval-ee-uzhasayuschey/
|
| 27 |
+
https://lenta.ru/news/2025/10/10/yak-130m/
|
| 28 |
+
https://lenta.ru/news/2025/10/10/rossiya-raskryla-otnoshenie-k-nobelevskoy-premii-mira-dlya-trampa/
|
| 29 |
+
https://lenta.ru/news/2025/10/10/amerikanets-rasskazal-o-mechte-stat-grazhdaninom-rossii/
|
| 30 |
+
https://lenta.ru/news/2025/10/10/putin-i-aliev-obnyalis-na-neformalnoy-vstreche/
|
| 31 |
+
https://lenta.ru/news/2025/10/10/viktoriya-bekhem-ob-yasnila-ugryumoe-vyrazhenie-litsa-na-krasnyh-dorozhkah/
|
| 32 |
+
https://lenta.ru/news/2025/10/10/rossiyanka-pobyvala-v-sochi-i-opisala-otdyh-na-kurorte-slovami-plyazhnyy-ad/
|
| 33 |
+
https://lenta.ru/news/2025/10/10/no-hands/
|
| 34 |
+
https://lenta.ru/news/2025/10/10/fon-der-lyayen-dali-shest-mesyatsev-na-ispravlenie-svoih-oshibok/
|
| 35 |
+
https://lenta.ru/news/2025/10/10/srednie-pensii-v-rossii-proseli-otnositelno-zarplat/
|
| 36 |
+
https://lenta.ru/news/2025/10/10/millionam-rossiyan-predskazali-uho-v-ten/
|
| 37 |
+
https://lenta.ru/news/2025/10/10/raskryt-podhod-trampa-k-uregulirovaniyu-konflikta-v-gaze/
|
| 38 |
+
https://lenta.ru/news/2025/10/10/v-moskve-ischez-svet/
|
| 39 |
+
https://lenta.ru/news/2025/10/10/na-ukraine-perekryli-dvizhenie-po-dambe-dneproges/
|
| 40 |
+
https://lenta.ru/news/2025/10/10/v-rossii-predrekli-otkaz-ssha-ot-otvetstvennosti-za-primenenie-raket-tomahawk/
|
| 41 |
+
https://lenta.ru/news/2025/10/10/v-dnepropetrovske-nachalsya-pozhar-posle-vzryvov/
|
| 42 |
+
https://lenta.ru/news/2025/10/10/vs-rossii-unichtozhili-obstrelivavshiy-belgorodskuyu-oblast-raschet-bpla/
|
| 43 |
+
https://lenta.ru/news/2025/10/09/v-rossii-izmenilis-tseny-na-zhilie/
|
| 44 |
+
https://lenta.ru/news/2025/10/10/vrach-otsenil-trend-na-uvelichenie-klitora/
|
| 45 |
+
https://lenta.ru/news/2025/10/10/amerikanskiy-polkovnik-zayavil-o-provale-podgotovki-evropy-k-voyne-s-rossiey/
|
| 46 |
+
https://lenta.ru/news/2025/10/10/v-gosdume-rasskazali-o-vliyanii-uchastnikov-svo-na-posleduyuschie-pokoleniya/
|
| 47 |
+
https://lenta.ru/news/2025/10/10/ssha-nanesli-serieznyy-uron-ekonomike-svoego-soyuznika/
|
| 48 |
+
https://lenta.ru/news/2025/10/10/otets-oblil-pertsovkoy-obidchikov-syna-vozle-shkoly-v-rossiyskom-regione/
|
| 49 |
+
https://lenta.ru/news/2025/10/10/u-eks-glavy-rossiyskoy-stroitelnoy-kompanii-nashli-osobuyu-cherno-beluyu-knigu/
|
| 50 |
+
https://lenta.ru/news/2025/10/10/starmer-vyskazalsya-o-sudbe-uvolennogo-iz-za-svyazey-s-epshteynom-posla-britanii/
|
| 51 |
+
https://lenta.ru/news/2025/10/10/partner-po-briks-narastil-postavki-myasa-rossii/
|
| 52 |
+
https://lenta.ru/news/2025/10/10/natsistskoe-nasledie-oslozhnilo-germanii-borbu-s-dronami/
|
| 53 |
+
https://lenta.ru/news/2025/10/10/rossiyskie-voyska-za-sutki-prodvinulis-v-dnr/
|
| 54 |
+
https://lenta.ru/news/2025/10/10/rossiyskiy-boets-rasskazal-o-prorytom-vsu-tunnele/
|
| 55 |
+
https://lenta.ru/news/2025/10/10/raskryta-shema-sokrytiya-nalogov-na-260-millionov-eks-glavoy-rossiyskoy-stroykompanii/
|
| 56 |
+
https://lenta.ru/news/2025/10/10/v-gosdume-zahoteli-zakonodatelno-ogranichit-chislo-koshek-i-sobak-v-kvartire/
|
| 57 |
+
https://lenta.ru/news/2025/10/10/v-rossii-rasskazali-o-vozmozhnoy-nezakonnoy-voenno-biologicheskoy-deyatelnosti-stran-nato/
|
| 58 |
+
https://lenta.ru/news/2025/10/10/voenkory-soobschili-o-novyh-udarah-po-kievu/
|
| 59 |
+
https://lenta.ru/news/2025/10/10/minprosvescheniya-ustanovilo-sroki-polucheniya-nachalnogo-obschego-obrazovaniya/
|
| 60 |
+
https://lenta.ru/news/2025/10/10/raskryty-roli-spetsposlannika-i-zyatya-trampa-v-uregulirovanii-konflikta-v-gaze/
|
| 61 |
+
https://lenta.ru/news/2025/10/10/perespavshaya-s-tysyachey-muzhchin-pornozvezda-bonni-blyu-otsenila-privlekatelnost-trampa/
|
| 62 |
+
https://lenta.ru/news/2025/10/10/v-rossii-otvetili-na-vopros-o-nobelevskoy-premii-mira-dlya-trampa/
|
| 63 |
+
https://lenta.ru/news/2025/10/10/na-ukraine-ivana-susanina-priznali-propagandoy-rossiyskogo-imperializma/
|
| 64 |
+
https://lenta.ru/news/2025/10/10/strana-briks-minimizirovala-postavki-kofe-v-rossiyu/
|
| 65 |
+
https://lenta.ru/news/2025/10/10/stalo-izvestno-o-planah-kieva-dostavit-patriot-v-harkov/
|
| 66 |
+
https://lenta.ru/news/2025/10/09/samaya-seksualnaya-sportsmenka-mira-pokazala-figuru-v-obtyagivayuschem-platie/
|
| 67 |
+
https://lenta.ru/news/2025/10/10/rossiyanka-ukrala-iz-elektrichki-sumku-s-dengami-i-popala-na-video/
|
| 68 |
+
https://lenta.ru/news/2025/10/10/moment-pogruzivshego-kiev-vo-tmu-udara-rakety-popal-na-video/
|
| 69 |
+
https://lenta.ru/news/2025/10/10/rossiya-popala-v-spisok-stran-s-samymi-deshevymi-yaytsami/
|
| 70 |
+
https://lenta.ru/news/2025/10/10/nazvany-tseli-nochnyh-udarov-po-kievu/
|
| 71 |
+
https://lenta.ru/news/2025/10/10/s-nachala-2025-goda-iz-vsu-dezertirovali-bolee-150-tysyach-chelovek/
|
| 72 |
+
https://lenta.ru/news/2025/10/10/raskryty-podrobnosti-o-nochnyh-atakah-vsu-na-rossiyu/
|
| 73 |
+
https://lenta.ru/news/2025/10/10/obnaruzhen-sposob-prokachat-immunitet-bez-lekarstv/
|
| 74 |
+
https://lenta.ru/news/2025/10/10/prodazham-kitayskih-legkovushek-predskazali-obrushenie/
|
| 75 |
+
https://lenta.ru/news/2025/10/10/zhitelya-tadzhikistana-ne-pustili-v-rossiyu-iz-za-video-o-terakte-v-krokuse/
|
| 76 |
+
https://lenta.ru/news/2025/10/10/v-partii-zelenskogo-zahoteli-snyat-klichko-s-posta-mera-kieva/
|
| 77 |
+
https://lenta.ru/news/2025/10/10/risk-vozniknoveniya-paniki-iz-za-tsen-na-sahar-otsenili/
|
| 78 |
+
https://lenta.ru/news/2025/10/10/v-rossii-otreagirovali-na-prizyv-evroparlamenta-po-rossiyskim-aktivam/
|
| 79 |
+
https://lenta.ru/news/2025/10/10/v-bundestage-rasskazali-o-glavnoy-pomehe-v-borbe-s-dronami-v-germanii/
|
| 80 |
+
https://lenta.ru/news/2025/10/10/pervyy-udar-gibrida-rakety-i-bomby-po-dnepru-snyali-na-video/
|
| 81 |
+
https://lenta.ru/news/2025/10/10/zemfira-sohranila-prava-na-svoi-pesni-v-rossii/
|
| 82 |
+
https://lenta.ru/news/2025/10/10/marochko-rasskazal-ob-otvode-komandovaniya-vsu-iz-severska/
|
| 83 |
+
https://lenta.ru/news/2025/10/10/energosistema-na-grani-kollapsa-armiya-rossii-vyvela-iz-stroya-klyuchevye-ges-i-tes-ukrainy-goroda-pogruzilis-vo-tmu/
|
| 84 |
+
https://lenta.ru/news/2025/10/10/kartavyh/
|
| 85 |
+
https://lenta.ru/news/2025/10/10/norvegii-predskazali-uzhestochenie-tarifov-iz-za-vozmozhnogo-otkaza-trampu-v-premii-mira/
|
| 86 |
+
https://lenta.ru/news/2025/10/10/deputat-rady-nazval-bessmyslennymi-prizyvy-ustroit-blekaut-v-rossii/
|
| 87 |
+
https://lenta.ru/news/2025/10/10/minzdrav-rasskazal-o-sostoyanii-vybroshennoy-ottsom-s-chetvertogo-etazha-trehletney-devochki/
|
| 88 |
+
https://lenta.ru/news/2025/10/10/dreyfovavshiy-dva-mesyatsa-v-ohotskom-more-rossiyanin-poydet-pod-sud/
|
| 89 |
+
https://lenta.ru/news/2025/10/10/sk-rasskazal-versiyu-propazhi-semi-v-krasnoyarskom-krae/
|
| 90 |
+
https://lenta.ru/news/2025/10/10/udar/
|
| 91 |
+
https://lenta.ru/news/2025/10/10/stalo-izvestno-o-bolshoy-strategii-trampa-v-otnoshenii-rossii/
|
| 92 |
+
https://lenta.ru/news/2025/10/10/devushka-vypila-bokal-vina-v-gostyah-u-kollegi-i-stala-prichinoy-ego-razvoda-s-zhenoy/
|
| 93 |
+
https://lenta.ru/news/2025/10/10/stoit-vykinut-tramp-predlozhil-isklyuchit-ispaniyu-iz-nato-kakaya-pretenziya-u-prezidenta-ssha-k-evropeyskoy-strane/
|
| 94 |
+
https://lenta.ru/news/2025/10/10/ushedshiy-iz-rossii-tehnologicheskiy-gigant-zaregistriroval-v-strane-tovarnyy-znak/
|
| 95 |
+
https://lenta.ru/news/2025/10/10/lyubovniki-zanyalis-seksom-v-tualete-dlya-invalidov-na-glazah-u-svideteley/
|
| 96 |
+
https://lenta.ru/news/2025/10/10/46-letnyaya-keyt-hadson-opublikovala-otkrovennoe-foto-pered-vyhodom-v-svet/
|
| 97 |
+
https://lenta.ru/news/2025/10/10/param-podskazali-neobychnyy-sposob-zanimatsya-seksom/
|
| 98 |
+
https://lenta.ru/news/2025/10/10/v-rossiyskom-gorode-sbivshego-lyudey-na-ostanovke-voditelya-zaderzhali/
|
| 99 |
+
https://lenta.ru/news/2025/10/10/marochko-soobschil-ob-okruzhenii-boytsov-vsu-v-rayone-stantsii-kupyansk/
|
| 100 |
+
https://lenta.ru/news/2025/10/10/poyavilis-podrobnosti-nochnogo-udara-po-ukraine/
|
| 101 |
+
https://lenta.ru/news/2025/10/10/gendirektora-monolitholdinga-abasova-zaderzhali/
|
| 102 |
+
https://lenta.ru/news/2025/10/10/byvshiy-analitik-tsru-rasskazal-o-zavershenii-konflikta-na-ukraine/
|
| 103 |
+
https://lenta.ru/news/2025/10/10/v-venesuele-nachali-podgotovku-k-vtorzheniyu-ssha/
|
| 104 |
+
https://lenta.ru/news/2025/10/10/vliyanie-vyvoda-iz-stroya-poloviny-gazodobychi-ukrainy-na-otopitelnyy-sezon-otsenili/
|
| 105 |
+
https://lenta.ru/news/2025/10/10/premiya/
|
| 106 |
+
https://lenta.ru/news/2025/10/10/ukrainskih-inzhenerov-ulichili-v-neznanii-yazyka/
|
| 107 |
+
https://lenta.ru/news/2025/10/10/u-beregov-filippin-zafiksirovali-esche-odno-zemletryasenie/
|
| 108 |
+
https://lenta.ru/news/2025/10/10/muzhchina-pozhalovalsya-na-prevraschayuschuyusya-v-yadovituyu-zmeyu-zhenu/
|
| 109 |
+
https://lenta.ru/news/2025/10/10/v-rossii-vyskazalis-o-gotovnosti-protivostoyat-vozmozhnomu-nastupleniyu-vsu/
|
| 110 |
+
https://lenta.ru/news/2025/10/10/vlyublennaya-para-poteryala-son-po-neozhidannoy-prichine/
|
| 111 |
+
https://lenta.ru/news/2025/10/10/v-ssha-raskryli-novuyu-tsel-rossii-posle-ukrainy/
|
| 112 |
+
https://lenta.ru/news/2025/10/10/levyy-bereg-kieva-ostalsya-bez-sveta/
|
| 113 |
+
https://lenta.ru/news/2025/10/10/rossiyanka-pobyvala-v-armenii-i-opisala-ee-slovami-prinyali-kak-rodnuyu/
|
| 114 |
+
https://lenta.ru/news/2025/10/10/rossiyane-brosilis-brat-mikrozaymy-na-iphone/
|
| 115 |
+
https://lenta.ru/news/2025/10/10/rossiyskie-drony-sorvali-rotatsiyu-vsu-v-zone-svo/
|
| 116 |
+
https://lenta.ru/news/2025/10/10/leopard-rasterzal-vosmiletnyuyu-devochku/
|
| 117 |
+
https://lenta.ru/news/2025/10/10/seksolog-oprovergla-suschestvovanie-klitoralnogo-orgazma/
|
| 118 |
+
https://lenta.ru/news/2025/10/10/zhurnalist-predrek-azhiotazh-na-kladbischah-ukrainy/
|
| 119 |
+
https://lenta.ru/news/2025/10/10/v-livane-nashli-primenenie-tunnelyam-hezbollah/
|
| 120 |
+
https://lenta.ru/news/2025/10/10/mer-tbilisi-rasskazal-o-predlozhenii-k-gruzii-nachat-voynu-rossiey-v-2022-godu/
|
| 121 |
+
https://lenta.ru/news/2025/10/10/kim-chen-yn-poobeschal-sdelat-severnuyu-koreyu-sotsialisticheskim-raem/
|
| 122 |
+
https://lenta.ru/news/2025/10/10/rossiyan-predupredili-o-zamaskirovannoy-pod-cheki-opasnosti-iz-whatsapp/
|
| 123 |
+
https://lenta.ru/news/2025/10/10/otkryt-sposob-opredelit-risk-raka-lyogkih-u-nekuryaschih/
|
| 124 |
+
https://lenta.ru/news/2025/10/10/v-rossii-s-1-marta-izmenitsya-krayniy-srok-oplaty-uslug-zhkh/
|
| 125 |
+
https://lenta.ru/news/2025/10/10/v-lnr-prizvali-ne-rasprostranyat-informatsiyu-s-sayta-mirotvorets/
|
| 126 |
+
https://lenta.ru/news/2025/10/10/genprokuror-nyu-yorka-obvinila-trampa-v-mesti/
|
| 127 |
+
https://lenta.ru/news/2025/10/10/kim-chen-yn-priobnyal-pevtsa-shaman/
|
| 128 |
+
https://lenta.ru/news/2025/10/10/v-ran-sprognozirovali-datu-sozhzheniya-zemli-solntsem/
|
| 129 |
+
https://lenta.ru/news/2025/10/10/krymskiy-most-vremenno-perekryli/
|
| 130 |
+
https://lenta.ru/news/2025/10/10/paren-zadal-nevinnyy-vopros-kollege-i-vyyasnil-pravdu-o-solenyh-ogurtsah/
|
| 131 |
+
https://lenta.ru/news/2025/10/10/detey-zhdal-zhutkiy-syurpriz-v-shkole-utrom/
|
| 132 |
+
https://lenta.ru/news/2025/10/10/obraz-haydi-klum-na-novyh-foto-opisali-v-seti-frazoy-stanovitsya-razvratnee/
|
| 133 |
+
https://lenta.ru/news/2025/10/10/v-kieve-zafiksirovali-pereboi-s-elektroenergiey/
|
| 134 |
+
https://lenta.ru/news/2025/10/10/izrail-odobril-plan-vozvrascheniya-iz-gazy-vseh-zalozhnikov/
|
| 135 |
+
https://lenta.ru/news/2025/10/10/amerikanskiy-polkovnik-vyskazalsya-o-voyne-s-rossiey/
|
| 136 |
+
https://lenta.ru/news/2025/10/10/blogersha-opisala-stranu-tsentralnoy-ameriki-slovami-produkty-dorozhe-chem-v-rossii/
|
| 137 |
+
https://lenta.ru/news/2025/10/10/uchitelnitsa-sovratila-12-letnego-malchika/
|
| 138 |
+
https://lenta.ru/news/2025/10/10/melaniya-tramp-proignorirovala-pismo-o-zverstvah-ukrainskih-vlastey/
|
| 139 |
+
https://lenta.ru/news/2025/10/10/na-ukraine-zayavili-o-pervom-primenenii-kaba-pri-udare-po-dnepropetrovsku/
|
| 140 |
+
https://lenta.ru/news/2025/10/10/rossiyan-predupredili-ob-uderzhanii-dolgov-po-nalogam-iz-zarplaty/
|
| 141 |
+
https://lenta.ru/news/2025/10/10/vozmozhnosti-trampa-otomstit-norvegii-za-neprisuzhdenie-nobelevskoy-premii-otsenili/
|
| 142 |
+
https://lenta.ru/news/2025/10/10/v-lipetskoy-oblasti-razbilsya-istrebitel-mig-31-chto-izvestno-o-proisshestvii/
|
| 143 |
+
https://lenta.ru/news/2025/10/10/sem-aftershokov-zafiksirovali-u-beregov-kamchatki/
|
| 144 |
+
https://lenta.ru/news/2025/10/10/v-mid-rossii-zayavili-o-prevraschenii-moldavii-v-bazu-snabzheniya-kieva/
|
| 145 |
+
https://lenta.ru/news/2025/10/10/boyfrend-reshil-ustroit-romantichnyy-syurpriz-devushke-i-edva-ne-poluchil-skovorodkoy/
|
| 146 |
+
https://lenta.ru/news/2025/10/10/durov-otpravil-trevozhnoe-soobschenie-v-den-svoego-rozhdeniya/
|
| 147 |
+
https://lenta.ru/news/2025/10/10/zablokirovannym-blogeram-razreshat-vernutsya-na-youtube/
|
| 148 |
+
https://lenta.ru/news/2025/10/10/v-ssha-nazvali-samuyu-opasnuyu-ugrozu-rossii-dlya-nato/
|
| 149 |
+
https://lenta.ru/news/2025/10/10/vyzhivshaya-zhenschina-rasskazala-o-napadenii-volka-lyudoeda/
|
| 150 |
+
https://lenta.ru/news/2025/10/10/seksolog-prizvala-zanimatsya-lyubovyu-pered-svidaniem/
|
| 151 |
+
https://lenta.ru/news/2025/10/10/vs-rossii-atakovali-energeticheskiy-ob-ekt-v-zaporozhie/
|
| 152 |
+
https://lenta.ru/news/2025/10/10/v-shveytsarii-prizvali-makrona-pokinut-post-prezidenta-frantsii/
|
| 153 |
+
https://lenta.ru/news/2025/10/10/voenkory-soobschili-o-massirovannoy-atake-ob-ektov-vsu-v-dnepropetrovske/
|
| 154 |
+
https://lenta.ru/news/2025/10/10/v-vengrii-rasskazali/
|
| 155 |
+
https://lenta.ru/news/2025/10/10/klichko-zayavil-o-vozmozhnyh-pereboyah-s-elektrichestvom-v-kieve/
|
| 156 |
+
https://lenta.ru/news/2025/10/10/stalo-izvestno-o-namerenii-vlastey-moldavii-snosit-pamyatniki-bortsam-s-fashizmom/
|
| 157 |
+
https://lenta.ru/news/2025/10/10/42-letnyaya-supermodel-ogolila-grud-na-kameru/
|
| 158 |
+
https://lenta.ru/news/2025/10/10/evroparlament-nazvali-souchastnikom-prestupleniy-zelenskogo/
|
| 159 |
+
https://lenta.ru/news/2025/10/10/klonirovavshiy-ovtsu-dolli-uchenyy-vnezapno-umer/
|
| 160 |
+
https://lenta.ru/news/2025/10/10/film-o-rassledovanii-hersha-podryva-severnyh-potokov-sobral-ovatsii/
|
| 161 |
+
https://lenta.ru/news/2025/10/10/byvshego-igroka-sbornoy-ukrainy-po-regbi-likvidirovali-v-kupyanske/
|
| 162 |
+
https://lenta.ru/news/2025/10/10/karpin-shutkoy-ob-izbienii-rasskazal-o-metodah-raboty-v-sbornoy-rossii/
|
| 163 |
+
https://lenta.ru/news/2025/10/10/v-gosdume-obsudili-borbu-s-zaytsem-so-zlobnoy-ulybkoy/
|
| 164 |
+
https://lenta.ru/news/2025/10/10/predstavitel-pevtsa-efrema-amiramova-podtverdila-smert-artista/
|
| 165 |
+
https://lenta.ru/news/2025/10/10/plemyannik-raspravilsya-s-dyadey-iz-za-karaoke/
|
| 166 |
+
https://lenta.ru/news/2025/10/10/predstavlen-effektivnyy-sposob-lecheniya-opasnyh-bolezney-mozga/
|
| 167 |
+
https://lenta.ru/news/2025/10/10/safonov-otsenil-otsutstvie-igrovoy-praktiki-v-pszh/
|
| 168 |
+
https://lenta.ru/news/2025/10/10/v-shveytsarii-prizvali-otmenit-status-dlya-ukrainskih-bezhentsev/
|
| 169 |
+
https://lenta.ru/news/2025/10/10/ssha-zahoteli-zapretit-aviakompaniyam-iz-kitaya-proletat-nad-rossiey/
|
| 170 |
+
https://lenta.ru/news/2025/10/10/voenbloger-rasskazal-ob-obnaruzhennom-v-kurskoy-oblasti-arsenale-vsu/
|
| 171 |
+
https://lenta.ru/news/2025/10/09/plakaty-s-ursuloy-fon-der-lyayen-v-obraze-kuklovoda-poyavilis-v-vengrii/
|
| 172 |
+
https://lenta.ru/news/2025/10/10/v-kieve-nachalis-pereboi-so-svetom/
|
| 173 |
+
https://lenta.ru/news/2025/10/10/10-oktyabrya-kakoy-prazdnik-segodnya-otmechayut-v-rossii-i-mire/
|
| 174 |
+
https://lenta.ru/news/2025/10/10/otek/
|
| 175 |
+
https://lenta.ru/news/2025/10/09/klebo-zayavil-o-nehvatke-konkurentsii-s-rossiyskimi-sportsmenami/
|
| 176 |
+
https://lenta.ru/news/2025/10/09/opublikovany-kadry-gigantskoy-avtomobilnoy-probki-v-kitae/
|
| 177 |
+
https://lenta.ru/news/2025/10/09/v-nebe-nad-ukrainoy-zametili-sotni-dronov/
|
| 178 |
+
https://lenta.ru/news/2025/10/09/poyavilis-podrobnosti-ob-upavshem-v-rossiyskom-regione-istrebitele/
|
| 179 |
+
https://lenta.ru/news/2025/10/09/tramp-soobschil-o-prodazhe-oruzhiya-nato-dlya-dalneyshey-otpravki-ukraine/
|
| 180 |
+
https://lenta.ru/news/2025/10/09/medvedya-zametili-za-poedaniem-soderzhimogo-musornyh-bakov/
|
| 181 |
+
https://lenta.ru/news/2025/10/09/v-seti-poyavilis-kadry-zapuska-bezekipazhnogo-katera-vsu/
|
| 182 |
+
https://lenta.ru/news/2025/10/09/istrebitel-razbilsya-v-rossiyskom-regione/
|
| 183 |
+
https://lenta.ru/news/2025/10/09/v-kieve-ob-yavili-vozdushnuyu-trevogu-iz-za-rossiyskih-geraney/
|
| 184 |
+
https://lenta.ru/news/2025/10/09/v-rade-raskritikovali-zayavlenie-syrskogo-o-kontrnastuplenii-vsu/
|
| 185 |
+
https://lenta.ru/news/2025/10/09/tramp-vyskazalsya-o/
|
| 186 |
+
https://lenta.ru/news/2025/10/09/tramp-zayavil-o-pokupke-ledokolov-u-finlyandii/
|
| 187 |
+
https://lenta.ru/news/2025/10/09/indoneziya-otkazalas-vydavat-vizy-gimnastam-iz-izrailya-dlya-uchastiya-v-chempionate-mira/
|
| 188 |
+
https://lenta.ru/news/2025/10/09/v-rade-zayavili-ob-uhudshenii-snabzheniya-vsu/
|
| 189 |
+
https://lenta.ru/news/2025/10/09/senat-ssha-vnov-otklonil-zakonoproekty-o-prodlenii-finansirovaniya/
|
| 190 |
+
https://lenta.ru/news/2025/10/09/tramp-predlozhil-vykinut-iz-nato-odnu-evropeyskuyu-stranu/
|
| 191 |
+
https://lenta.ru/news/2025/10/09/tramp-zayavil-o-namerenii/
|
| 192 |
+
https://lenta.ru/news/2025/10/09/tramp-otsenil-ugrozu-finlyandii-so-storony-rossii/
|
| 193 |
+
https://lenta.ru/news/2025/10/09/tramp-vyskazalsya-o-peregovorah-rossii-i-ukrainy/
|
| 194 |
+
https://lenta.ru/news/2025/10/09/frantsuzskiy-politik-raskritikoval-slova-zelenskogo-o-peredache-kivu-raket/
|
| 195 |
+
https://lenta.ru/news/2025/10/09/ubytki-vsu-iz-za-poteri-shesti-sau-otsenili/
|
| 196 |
+
https://lenta.ru/news/2025/10/09/vremennye-ogranicheniya-vveli-v-aeroportu-rossiyskogo-goroda/
|
| 197 |
+
https://lenta.ru/news/2025/10/09/poyavilis-kadry-oblomkov-noveyshey-ukrainskoy-rakety-flamingo/
|
| 198 |
+
https://lenta.ru/news/2025/10/09/polskie-pogranichniki-zaderzhali-hotevshego-popast-v-es-podrostka/
|
| 199 |
+
https://lenta.ru/news/2025/10/09/na-ukraine-priznali-prevoshodstvo-rossii-v-proizvodstve-dronov/
|
| 200 |
+
https://lenta.ru/news/2025/10/09/aktera-zbrueva-pereveli-v-drugoe-meduchrezhdenie-posle-gospitalizatsii/
|
| 201 |
+
https://lenta.ru/news/2025/10/09/v-belgii-predotvratili-pokushenie-na-politikov-s-pomoschyu-bpla/
|
| 202 |
+
https://lenta.ru/news/2025/10/09/samaya-seksualnaya-v-mire-sudya-snyalas-v-kozhanom-mikrobikini/
|
| 203 |
+
https://lenta.ru/news/2025/10/09/borba-za-gorod-v-zone-svo-zakonchilas-podnyatiem-rossiyskogo-flaga/
|
| 204 |
+
https://lenta.ru/news/2025/10/09/hamas-podtverdil-zavershenie-voyny-s-izrailem/
|
| 205 |
+
https://lenta.ru/news/2025/10/09/na-zapade-rasskazali-o-podgotovke-armii-ssha/
|
| 206 |
+
https://lenta.ru/news/2025/10/09/tramp-ob-yavil-o-skorom-podpisanii-dogovorennosti-ob-uregulirovanii-v-gaze/
|
| 207 |
+
https://lenta.ru/news/2025/10/09/tramp-zadumalsya-ob-okazanii-uslugi-kitayu-za-trillion-dollarov/
|
| 208 |
+
https://lenta.ru/news/2025/10/09/voenkor-soobschil-o-planah-kieva-ustroit-blekaut-v-belgorode/
|
| 209 |
+
https://lenta.ru/news/2025/10/09/ssha-predrekli-otsutstvie-sredstv-na-pereosnaschenie-yadernogo-arsenala/
|
| 210 |
+
https://lenta.ru/news/2025/10/09/baronet/
|
| 211 |
+
https://lenta.ru/news/2025/10/09/byvshiy-ministr-sporta-rossii-otreagiroval-na-slova-osudivshego-uchastie-rossiyan-v-nhl-gasheka/
|
| 212 |
+
https://lenta.ru/news/2025/10/09/samaya-izvestnaya-pevitsa-v-mire-spela-o-penise-zheniha/
|
| 213 |
+
https://lenta.ru/news/2025/10/09/ob-yasnena-polza-koshek-dlya-zdorovya-serdtsa/
|
| 214 |
+
https://lenta.ru/news/2025/10/09/chervinskiy/
|
| 215 |
+
https://lenta.ru/news/2025/10/09/vuchich-lishilsya-vozmozhnosti-priehat-moskvu-iz-za-odnoy-zhenschiny/
|
| 216 |
+
https://lenta.ru/news/2025/10/09/moskva-azal/
|
| 217 |
+
https://lenta.ru/news/2025/10/09/glava-ofisa-zelenskogo-prigrozil-razrushit-infrastrukturu-rossii/
|
| 218 |
+
https://lenta.ru/news/2025/10/09/izvestnaya-rossiyskaya-televeduschaya-rasskazala-ob-aborte-i-nazvala-ego-muchitelnym-resheniem/
|
| 219 |
+
https://lenta.ru/news/2025/10/09/rossiyan-predupredili-ob-obshirnoy-tsiklonicheskoy-depressii/
|
| 220 |
+
https://lenta.ru/news/2025/10/09/byvshaya-zhena-pashi-tehnika-obratilas-k-reperu-vacio/
|
| 221 |
+
https://lenta.ru/news/2025/10/09/stalo-izvestno-sostoyanie-postradavshih-posle-naezda-avtomobilya-na-ostanovku-v-novosibirske/
|
| 222 |
+
https://lenta.ru/news/2025/10/09/putin-sozvonilsya-s-premierom-odnoy-strany/
|
| 223 |
+
https://lenta.ru/news/2025/10/09/42-letnyaya-zvezda-univera-pokazala-foto-v-pidzhake-na-goloe-telo/
|
| 224 |
+
https://lenta.ru/news/2025/10/09/znatoku-rossiyskoy-kuhni-izbrali-meru-presecheniya-v-moskve/
|
| 225 |
+
https://lenta.ru/news/2025/10/09/vrach-predlozhil-uluchshayuschuyu-psihicheskoe-sostoyanie-dietu/
|
| 226 |
+
https://lenta.ru/news/2025/10/09/evropeyskuyu-stranu-ulichili-v-opasnom-dlya-vsey-planety-deystvii/
|
| 227 |
+
https://lenta.ru/news/2025/10/09/krupneyshaya-ekonomika-evropy-stolknulas-s-usugubleniem-krizisa/
|
| 228 |
+
https://lenta.ru/news/2025/10/09/moskva-baku/
|
| 229 |
+
https://lenta.ru/news/2025/10/09/v-ssha-nazvali-klyuchevoy-faktor-v-prekraschenii-konflikta-na-ukraine/
|
| 230 |
+
https://lenta.ru/news/2025/10/09/v-kremle-rasskazali-ob-ostryh-problemah-v-razgovore-putina-i-alieva/
|
| 231 |
+
https://lenta.ru/news/2025/10/09/pereehavshiy-iz-rossii-v-izrail-izvestnyy-komik-nazval-izmenivshiy-ego-sudbu-den/
|
| 232 |
+
https://lenta.ru/news/2025/10/09/dzhuliya-foks-ogolilas-na-kovrovoy-dorozhke/
|
| 233 |
+
https://lenta.ru/news/2025/10/09/air/
|
| 234 |
+
https://lenta.ru/news/2025/10/09/doch-eks-glavy-verhovnogo-suda-rossiyskogo-regiona-svyazali-s-kriminalom/
|
| 235 |
+
https://lenta.ru/news/2025/10/09/poyavilos-video-s-mesta-dtp-na-avtobusnoy-ostanovke-v-rossiyskom-gorode/
|
| 236 |
+
https://lenta.ru/news/2025/10/09/turist-prygnul-s-balkona-pri-popytke-spastis-ot-pozhara-v-otele-i-ne-vyzhil/
|
| 237 |
+
https://lenta.ru/news/2025/10/09/rybolov-poymal-ogromnogo-soma-dyavola/
|
| 238 |
+
https://lenta.ru/news/2025/10/09/vsu-podorvali-ammiakoprovod-zachem-im-eto-bylo-nuzhno-i-chto-izvestno-o-posledstviyah/
|
| 239 |
+
https://lenta.ru/news/2025/10/09/na-ukraine-obratili-vnimanie-na-glavnuyu-voennuyu-problemu-kieva/
|
| 240 |
+
https://lenta.ru/news/2025/10/09/turisticheskiy-avtobus-s-passazhirami-iz-rossii-popal-v-dtp-v-abhazii/
|
| 241 |
+
https://lenta.ru/news/2025/10/09/tramp-vyskazalsya-ob-uregulirovanii-na-ukraine-posle-sdelki-po-gaze/
|
| 242 |
+
https://lenta.ru/news/2025/10/09/59-letnyaya-ekaterina-andreeva-pokazala-na-kameru-otkrovennoe-dekolte/
|
| 243 |
+
https://lenta.ru/news/2025/10/09/nazvana-prichina-obsuzhdeniya-putinym-i-alievym-krusheniya-samoleta-azal/
|
| 244 |
+
https://lenta.ru/news/2025/10/09/vosmiletnyaya-devochka-popala-pod-udar-ukrainskogo-drona-v-rossiyskom-regione/
|
| 245 |
+
https://lenta.ru/news/2025/10/09/prodazhi-odnogo-vida-mashin-v-rossii-obvalilis/
|
| 246 |
+
https://lenta.ru/news/2025/10/09/dva-moschnyh-zemletryaseniya-proizoshli-v-rossiyskom-regione/
|
| 247 |
+
https://lenta.ru/news/2025/10/09/orel-sbil-dron-v-kavkazskom-regione-rossii-i-popal-na-video/
|
| 248 |
+
https://lenta.ru/news/2025/10/09/geroy-intervyu-sobchak-otreagiroval-na-udalenie-video-iz-za-lgbt-propagandy/
|
| 249 |
+
https://lenta.ru/news/2025/10/09/sadovnik-podstrigal-kusty-i-poluchil-smertelnyy-udar-tokom/
|
| 250 |
+
https://lenta.ru/news/2025/10/09/nad-bazoy-nato-proletel/
|
| 251 |
+
https://lenta.ru/news/2025/10/09/stalo-izvestno-o-pervom-udare-kab-po-dnepropetrovsku/
|
| 252 |
+
https://lenta.ru/news/2025/10/09/zelenskiy-vyskazalsya-ob-udarah-po-leningradskoy-oblasti/
|
| 253 |
+
https://lenta.ru/news/2025/10/09/vrach-raskryl-neozhidannye-prichiny-postoyannoy-boli-v-kolene/
|
| 254 |
+
https://lenta.ru/news/2025/10/09/v-rossiyskom-gorode-avtomobil-protaranil-ostanovku-s-lyudmi/
|
| 255 |
+
https://lenta.ru/news/2025/10/09/poyavilis-podrobnosti-o-moskovskom-nasilnike/
|
| 256 |
+
https://lenta.ru/news/2025/10/09/v-gosdume-rasskazali-o-proverke-vyskazyvaniy-klaudino-ob-svo/
|
| 257 |
+
https://lenta.ru/news/2025/10/09/v-gosdume-vyskazalis-o-vnedrenii-kokoshnika-v-shkolnuyu-formu/
|
| 258 |
+
https://lenta.ru/news/2025/10/09/usloviya-zelenskogo-po-vydvizheniyu-trampa-na-nobelevskuyu-premiyu-otsenili/
|
| 259 |
+
https://lenta.ru/news/2025/10/09/rossiyskie-soldaty-pokazali-naydennyy-v-pakete-s-ostankami-boytsa-vsu-pod-kurskom-predmet/
|
| 260 |
+
https://lenta.ru/news/2025/10/09/dvigatel-passazhirskogo-samoleta-zagorelsya-na-vzletnoy-polose-i-popal-na-video/
|
| 261 |
+
https://lenta.ru/news/2025/10/09/silnyy-pozhar-v-mnogoetazhke-v-rossiyskom-gorode-millionnike-popal-na-video/
|
| 262 |
+
https://lenta.ru/news/2025/10/09/vyyavlena-unikalnaya-polza-privychnoy-yagody/
|
| 263 |
+
https://lenta.ru/news/2025/10/09/tualet-s-razvlecheniyami-otkryli-v-moskve/
|
| 264 |
+
https://lenta.ru/news/2025/10/09/voenkor-rasskazal-o-serieznyh-pereboyah-na-zheleznyh-dorogah-ukrainy-posle-udarov-vs-rossii/
|
| 265 |
+
https://lenta.ru/news/2025/10/09/tokaev-anonsiroval-svoy-vizit-v-rossiyu/
|
| 266 |
+
https://lenta.ru/news/2025/10/09/medvedev-poblagodaril-kim-chen-yna/
|
| 267 |
+
https://lenta.ru/news/2025/10/09/proizvodstvo-repressirovannogo-v-1930-h-godah-agittekstilya-vozobnovili-v-rossii/
|
| 268 |
+
https://lenta.ru/news/2025/10/09/na-ukraine-nazvali-tsel-udarov-po-zheleznodorozhnoy-infrastrukture/
|
| 269 |
+
https://lenta.ru/news/2025/10/09/razvod-samoylovoy-s-dzhiganom-oprovergli/
|
| 270 |
+
https://lenta.ru/news/2025/10/09/minoborony-rossii-obvinilo-vsu-v-podryve-ammiakoprovoda-v-zone-svo-pri-otstuplenii/
|
| 271 |
+
https://lenta.ru/news/2025/10/09/poplavskaya-zayavila-o-propityh-mozgah-lazarevoy/
|
| 272 |
+
https://lenta.ru/news/2025/10/09/rabotnika-zavoda-razdavilo-betonomeshalkoy/
|
| 273 |
+
https://lenta.ru/news/2025/10/09/rossiyskiy-shkolnik-udaril-beremennuyu-uchitelnitsu/
|
| 274 |
+
https://lenta.ru/news/2025/10/09/tsb-ponizil-kurs-dollara/
|
| 275 |
+
https://lenta.ru/news/2025/10/09/raskryty-detali-razdela-roskoshnogo-imuschestva-samoylovoy-i-dzhigana/
|
| 276 |
+
https://lenta.ru/news/2025/10/09/v-rossii-uvelichilos-kolichestvo-kiberatak-na-kriticheskuyu-informatsionnuyu-infrastrukturu/
|
| 277 |
+
https://lenta.ru/news/2025/10/09/v-podmoskovie-propal-lenin/
|
| 278 |
+
https://lenta.ru/news/2025/10/09/v-mid-rossii-rasskazali-o-pomoschi-kurskoy-oblasti-so-storony-serbii/
|
| 279 |
+
https://lenta.ru/news/2025/10/09/nazvany-pyat-prichin-chastyh-prostud/
|
| 280 |
+
https://lenta.ru/news/2025/10/09/vnuk-zhestoko-raspravilsya-s-babushkoy-v-rossiyskom-regione/
|
| 281 |
+
https://lenta.ru/news/2025/10/09/sedokova-poblagodarila-repera-macan-za-lyubov/
|
| 282 |
+
https://lenta.ru/news/2025/10/09/skabeeva-otreagirovala-na-slova-zelenskogo-o-tomahawk/
|
| 283 |
+
https://lenta.ru/news/2025/10/09/zvezda-garri-pottera-raskryla-pravdu-o-koltse-na-bezymyannom-paltse-s-nedeli-mody/
|
| 284 |
+
https://lenta.ru/news/2025/10/09/s-doprosom-vybrosivshego-doch-iz-okna-rossiyanina-voznikli-slozhnosti/
|
| 285 |
+
https://lenta.ru/news/2025/10/09/nazvana-data-pervogo-snega-v-moskve/
|
| 286 |
+
https://lenta.ru/news/2025/10/09/poyavilis-podrobnosti-o-zazhivo-sgorevshem-na-parkovke-v-podmoskovie-veterane-svo/
|
| 287 |
+
https://lenta.ru/news/2025/10/09/italyanskih-plovchih-otstranili-ot-chempionata-evropy-za-krazhu/
|
| 288 |
+
https://lenta.ru/news/2025/10/09/opublikovano-video-s-rastaptyvayuschim-golovu-passazhirki-tramvaya-rossiyaninym/
|
| 289 |
+
https://lenta.ru/news/2025/10/09/otpravivsheysya-v-pohod-devushke-prishlos-spasatsya-begstvom-ot-gologo-muzhchiny/
|
| 290 |
+
https://lenta.ru/news/2025/10/09/odnomu-segmentu-avtorynka-rossii-sprognozirovali-ubiystvo-iz-za-novogo-utilya/
|
| 291 |
+
https://lenta.ru/news/2025/10/09/putin-nazval-sposob-razresheniya-voprosa-vokrug-iranskoy-yadernoy-programmy/
|
| 292 |
+
https://lenta.ru/news/2025/10/09/evropu-prizvali-vnedrit-ukrainskiy-sposob-borby-s-rossiey/
|
| 293 |
+
https://lenta.ru/news/2025/10/09/dzhennifer-eniston-raskryla-prichinu-otsutstviya-detey/
|
| 294 |
+
https://lenta.ru/news/2025/10/09/deputat-rady-uvidel-rol-polshi-v-podryve-severnogo-potoka-2/
|
| 295 |
+
https://lenta.ru/news/2025/10/09/osuzhdennyy-rossiyskiy-student-priznalsya-v-gosizmene/
|
| 296 |
+
https://lenta.ru/news/2025/10/09/vneshnost-zvezdy-klipa-freestyler-na-novom-foto-udivila-russkoyazychnuyu-publiku/
|
| 297 |
+
https://lenta.ru/news/2025/10/09/durianovyy-millioner-lishil-syna-deneg-za-ego-izmenu/
|
| 298 |
+
https://lenta.ru/news/2025/10/09/nobelevskuyu-premiyu-po-literature-poluchil-avtor-sataninskogo-tango/
|
| 299 |
+
https://lenta.ru/news/2025/10/09/evropu-predupredili-ob-opasnosti-fon-der-lyayen/
|
| 300 |
+
https://lenta.ru/news/2025/10/09/v-rossii-kafe-zakryli-iz-za-deystvuyuschego-zapreta-na-rabotu-migrantov/
|
| 301 |
+
https://lenta.ru/news/2025/10/09/na-solntse-proizoshla-moschnaya-vspyshka-i-popala-na-foto/
|
| 302 |
+
https://lenta.ru/news/2025/10/09/glava-gazproma-poobeschal-rossii-anomalno-holodnuyu-zimu/
|
| 303 |
+
https://lenta.ru/news/2025/10/09/zelenskiy-nazval-usloviya-kieva-dlya-vydvizheniya-trampa-na-nobelevskuyu-premiyu-mira/
|
| 304 |
+
https://lenta.ru/news/2025/10/09/tarasova-zastupilas-za-smenivshih-grazhdanstvo-rossiyskih-sportsmenov/
|
| 305 |
+
https://lenta.ru/news/2025/10/09/predskazano-buduschee-tsen-na-zhilie-v-moskve/
|
| 306 |
+
https://lenta.ru/news/2025/10/09/polsha-zapodozrila-sotrudnika-zagsa-v-shpionade-v-polshu-rossiyu/
|
| 307 |
+
https://lenta.ru/news/2025/10/09/v-evropeyskoy-strane-zapustili-prodazhi-kultovogo-rossiyskogo-vnedorozhnika/
|
| 308 |
+
https://lenta.ru/news/2025/10/09/ob-yavleno-o-pervom-unichtozhenii-noveyshey-rakety-vsu-na-skorosti-v-600-kilometrov-v-chas/
|
| 309 |
+
https://lenta.ru/news/2025/10/09/perechisleny-trebuyuschie-nemedlennogo-vyzova-skoroy-pomoschi-simptomy/
|
| 310 |
+
https://lenta.ru/news/2025/10/09/rossiyan-predupredili-ob-opasnoy-rassylke-v-internete/
|
| 311 |
+
https://lenta.ru/news/2025/10/09/viktoriya-bekhem-opisala-zhizn-s-rpp-frazoy-ya-nachala-ne-lyubit-sebya/
|
| 312 |
+
https://lenta.ru/news/2025/10/09/rossiyanam-nazvali-srok-perehoda-na-zimnyuyu-rezinu/
|
| 313 |
+
https://lenta.ru/news/2025/10/09/poyavilos-foto-doma-na-kolesah-sgorevshego-zazhivo-veterana-svo/
|
| 314 |
+
https://lenta.ru/news/2025/10/09/letevshiy-v-turtsiyu-inostrannyy-samolet-sel-v-rossii-iz-za-bolezni-passazhirov/
|
| 315 |
+
https://lenta.ru/news/2025/10/09/rossiyanka-dva-goda-poluchala-pensiyu-za-dyadyu-posle-ego-smerti/
|
| 316 |
+
https://lenta.ru/news/2025/10/09/shufutinskiy-otvetil-na-soobscheniya-ob-onkologii-frazoy-ne-dozhdetes/
|
| 317 |
+
https://lenta.ru/news/2025/10/09/zhiteley-slavyanska-prizvali-k-evakuatsii/
|
| 318 |
+
https://lenta.ru/news/2025/10/09/vietnam-flood/
|
| 319 |
+
https://lenta.ru/news/2025/10/09/putin-otreagiroval-na-initsiativy-trampa-po-gaze/
|
| 320 |
+
https://lenta.ru/news/2025/10/09/zhiteli-rossiyskogo-goroda-pozhalovalis-na-stayu-neulovimyh-sobak/
|
| 321 |
+
https://lenta.ru/news/2025/10/09/doch-bandita-tsapka-oshtrafovali-za-pyanoe-dtp/
|
| 322 |
+
https://lenta.ru/news/2025/10/09/rossiyskie-voennye-likvidirovali-amerikanskogo-naemnika-iz-azova/
|
| 323 |
+
https://lenta.ru/news/2025/10/09/v-peterburge-peredumali-ubirat-ulitsy-robotami/
|
| 324 |
+
https://lenta.ru/news/2025/10/09/vuchich-porassuzhdal-o-vine-rossii-za-sanktsii-ssha/
|
| 325 |
+
https://lenta.ru/news/2025/10/09/new-shepard-sovershil-shestoy-pilotiruemyy-polet-v-2025-godu/
|
| 326 |
+
https://lenta.ru/news/2025/10/09/u-pokinuvshego-rossiyu-proizvoditelya-gazirovki-upala-pribyl/
|
| 327 |
+
https://lenta.ru/news/2025/10/09/muzhchina-s-el-tri-kilogramma-marmelada-za-tri-dnya-i-popal-v-bolnitsu/
|
| 328 |
+
https://lenta.ru/news/2025/10/09/iz-rossiyskogo-vuza-evakuirovali-neskolko-soten-studentov/
|
| 329 |
+
https://lenta.ru/news/2025/10/09/inostrannyy-lider-pozdravil-putina-na-russkom-yazyke/
|
| 330 |
+
https://lenta.ru/news/2025/10/09/ubezhdavshie-shkolnikov-oformlyat-karty-dlya-moshennikov-rossiyane-stali-figurantami-dela/
|
| 331 |
+
https://lenta.ru/news/2025/10/09/anita-tsoy-rasskazala-o-borbe-s-depressiey/
|
| 332 |
+
https://lenta.ru/news/2025/10/09/nazvany-lyubimye-strany-rossiyskih-millenialov/
|
| 333 |
+
https://lenta.ru/news/2025/10/09/nayden-sposob-mgnovenno-opredelit-tuberkulez-bez-analizov/
|
| 334 |
+
https://lenta.ru/news/2025/10/09/ukrainskaya-nalogovaya-sluzhba-nachala-pisat-onlifanschitsam/
|
| 335 |
+
https://lenta.ru/news/2025/10/09/jan-kholod/
|
| 336 |
+
https://lenta.ru/news/2025/10/09/stilisty-nazvali-snimayuschie-ustalost-tsveta-v-odezhde/
|
| 337 |
+
https://lenta.ru/news/2025/10/09/trener-vashingtona-otsenil-igru-ovechkina-v-pervom-matche-novogo-sezona-nhl/
|
| 338 |
+
https://lenta.ru/news/2025/10/09/izvestnuyu-lgbt-aktivistku-vnesli-v-reestr-terroristov-v-rossii/
|
| 339 |
+
https://lenta.ru/news/2025/10/09/raskryta-summa-vyplat-uchastnikam-svo-i-ih-semyam/
|
| 340 |
+
https://lenta.ru/news/2025/10/09/proniknovenie-fanata-v-kvartiru-rossiyskoy-pevitsy-nazvali-piar-hodom/
|
| 341 |
+
https://lenta.ru/news/2025/10/09/samoylova-razvedetsya-s-dzhiganom/
|
| 342 |
+
https://lenta.ru/news/2025/10/09/predskazano-vliyanie-vstrechi-putina-i-alieva-na-otnosheniya-rossii-i-azerbaydzhana/
|
| 343 |
+
https://lenta.ru/news/2025/10/09/v-mahachkale-izbrali-novogo-mera/
|
| 344 |
+
https://lenta.ru/news/2025/10/09/rezultativnost-rossiyskoy-zanozy-v-zone-svo-raskryli/
|
| 345 |
+
https://lenta.ru/news/2025/10/09/v-rossiyskom-gorode-vybrali-novogo-mera/
|
| 346 |
+
https://lenta.ru/news/2025/10/09/msk-zima/
|
| 347 |
+
https://lenta.ru/news/2025/10/09/uitaker-rasskazal-o-esche-ne-peredannom-ukraine-moschnom-oruzhii-ssha/
|
| 348 |
+
https://lenta.ru/news/2025/10/09/vstrecha-putina-i-alieva-zavershilas/
|
| 349 |
+
https://lenta.ru/news/2025/10/09/kandidat-v-premiery-chehii-babish-anonsiroval-vizit-na-ukrainu/
|
| 350 |
+
https://lenta.ru/news/2025/10/09/okeanolog-usomnilsya-v-dostovernosti-prognozov-ob-izmeneniyah-v-mirovom-okeane/
|
| 351 |
+
https://lenta.ru/news/2025/10/09/arheologi-nashli-redkiy-kamennyy-molot-vozrastom-devyat-tysyach-let/
|
| 352 |
+
https://lenta.ru/news/2025/10/09/mishustin-rasskazal-o-posobii-dlya-samozanyatyh/
|
| 353 |
+
https://lenta.ru/news/2025/10/09/v-moskovskom-kafe-muzhchina-s-lopatoy-pohitil-konstruktor-dlya-syna-i-popal-na-video/
|
| 354 |
+
https://lenta.ru/news/2025/10/09/babkina-posovetovala-rossiyanam-borotsya-s-osenney-handroy-uborkoy/
|
| 355 |
+
https://lenta.ru/news/2025/10/09/nazvana-bezopasnaya-sutochnaya-doza-kofe/
|
| 356 |
+
https://lenta.ru/news/2025/10/09/nachalnitsu-sudebnyh-pristavov-etapiruyut-v-moskvu/
|
| 357 |
+
https://lenta.ru/news/2025/10/09/stalo-izvestno-o-nedvizhimosti-bulanovoy-v-moskve-i-peterburge/
|
| 358 |
+
https://lenta.ru/news/2025/10/09/msk-dozhdi/
|
| 359 |
+
https://lenta.ru/news/2025/10/09/v-rossiyskom-regione-zhiteli-pozhalovalis-na-razryvayuschih-domashnih-zhivotnyh-sobak/
|
| 360 |
+
https://lenta.ru/news/2025/10/09/iphone-fold/
|
| 361 |
+
https://lenta.ru/news/2025/10/09/eks-premier-ukrainy-predrek-territorialnoe-buduschee-strany/
|
| 362 |
+
https://lenta.ru/news/2025/10/09/zakon-o-shtrafah-za-poisk-ekstremistskih-materialov-mogut-izmenit/
|
| 363 |
+
https://lenta.ru/news/2025/10/09/mnogodetnaya-mat-iz-moskvy-podderzhala-vsu-i-popala-v-sizo/
|
| 364 |
+
https://lenta.ru/news/2025/10/09/polzovateli-max-smogut-sozdat-tsifrovoy-id-po-voditelskomu-udostovereniyu/
|
| 365 |
+
https://lenta.ru/news/2025/10/09/moskovskiy-dvornik-podralsya-s-derevom-i-popal-na-video/
|
| 366 |
+
https://lenta.ru/news/2025/10/09/umer-zvezda-shansona-efrem-amiramov-ispolnitelyu-hita-molodaya-bylo-69-let/
|
| 367 |
+
https://lenta.ru/news/2025/10/09/v-sovfede-nazvali-realnogo-sponsora-terrorizma/
|
| 368 |
+
https://lenta.ru/news/2025/10/09/sanktsii-ssha-udarili-po-postavkam-nefti-v-serbiyu/
|
| 369 |
+
https://lenta.ru/news/2025/10/09/v-rossii-desyatki-krepkih-muzhchin-v-chernom-v-balaklavah-i-s-nashivkami-vnov-vyshli-na-ulitsy/
|
| 370 |
+
https://lenta.ru/news/2025/10/09/stalo-izvestno-o-smerti-rossiyanina-na-polu-v-tualete-bolnitsy/
|
| 371 |
+
https://lenta.ru/news/2025/10/09/bitva/
|
| 372 |
+
https://lenta.ru/news/2025/10/09/sobirayuschimsya-v-kitay-rossiyanam-dali-odin-sovet/
|
| 373 |
+
https://lenta.ru/news/2025/10/09/postavschikam-kompleksov-razminirovaniya-dlya-rosgvardii-nazvali-srok/
|
| 374 |
+
https://lenta.ru/news/2025/10/09/v-rossii-zayavili-o-silnom-otstavanii-v-kosmonavtike/
|
| 375 |
+
https://lenta.ru/news/2025/10/09/byvshaya-zhena-mamaeva-proigrala-v-borbe-za-elitnye-apartamenty-eks-muzha/
|
| 376 |
+
https://lenta.ru/news/2025/10/09/zelenskiy-pogovoril-s-kritikovavshim-pomosch-kievu-pobeditelem-vyborov-v-chehii/
|
| 377 |
+
https://lenta.ru/news/2025/10/09/guzeeva-otvetila-na-posvyaschennyy-ey-vypusk-programmy-kudryavtsevoy-i-sochla-ee-nezdorovoy/
|
| 378 |
+
https://lenta.ru/news/2025/10/09/tur-rossiyskoy-repershi-okazalsya-pod-ugrozoy-iz-za-propagandy-lgbt/
|
| 379 |
+
https://lenta.ru/news/2025/10/09/novuyu-strizhku-zheny-dzhordzha-kluni-podnyali-na-smeh-v-seti/
|
| 380 |
+
https://lenta.ru/news/2025/10/09/ukhodi/
|
| 381 |
+
https://lenta.ru/news/2025/10/09/rukovodstvo-ob-yasnilos-posle-organizatsii-dnya-beremennyh-v-rossiyskom-detsadu/
|
| 382 |
+
https://lenta.ru/news/2025/10/09/iz-kanalizatsii-odnogo-goroda-izvlekli-zhirberg-vesom-100-tonn/
|
| 383 |
+
https://lenta.ru/news/2025/10/09/skoruyu-vyzvali-v-sud-posle-slov-prokurora-po-delu-ob-obmane-rosgvardii-na-283-milliona/
|
| 384 |
+
https://lenta.ru/news/2025/10/09/evroparlament-prizval-nemedlenno-konfiskovat-rossiyskie-aktivy/
|
| 385 |
+
https://lenta.ru/news/2025/10/09/putin-raskryl-novuyu-detal-v-krushenii-samoleta-azal/
|
| 386 |
+
https://lenta.ru/news/2025/10/09/putin-raskryl-alievu-podrobnosti-krusheniya-samoleta-azal-iz-za-chego-proizoshla-aviakatastrofa/
|
| 387 |
+
https://lenta.ru/news/2025/10/09/moskvichey-prizvali-ne-boyatsya-letuchih-myshey/
|
| 388 |
+
https://lenta.ru/news/2025/10/09/aliev-ob-yavil-uspehe-otnosheniy-s-rossiey-v-2025-godu/
|
| 389 |
+
https://lenta.ru/news/2025/10/09/nazvana-vozmozhnaya-prichina-pozhara-v-dome-na-kolesah-sgorevshego-zazhivo-veterana-svo/
|
| 390 |
+
https://lenta.ru/news/2025/10/10/zhenschina-s-chetvertoy-stadiey-raka-rasskazala-o-neochevidnyh-simptomah-zabolevaniya/
|
| 391 |
+
https://lenta.ru/news/2025/10/10/rossiyskaya-aviakompaniya-narushila-prava-passazhirov-i-poplatilas/
|
| 392 |
+
https://lenta.ru/news/2025/10/10/stali-izvestny-plany-ssha-posle-pervogo-etapa-mirnogo-plana-po-gaze/
|
| 393 |
+
https://lenta.ru/news/2025/10/10/vengriya-prizvala-vseh-zapomnit-odin-vazhnyy-urok/
|
| 394 |
+
https://lenta.ru/news/2025/10/10/gospitalizirovan-byvshiy-sudya-verhovnogo-suda-rossii-momotov/
|
| 395 |
+
https://lenta.ru/news/2025/10/10/rossiyanka-nashla-syna-podrostka-v-komnate-mertvym-i-s-ozhogom-na-grudi/
|
| 396 |
+
https://lenta.ru/news/2025/10/10/v-krymu-vskryli-peredavavshego-informatsiyu-o-korablyah-rossiyanina/
|
| 397 |
+
https://lenta.ru/news/2025/10/10/peskov-obvinil-kiev-v-ostanovke-peregovornogo-protsessa/
|
| 398 |
+
https://lenta.ru/news/2025/10/10/gendirektor-dtek-zayavil-o-nesposobnosti-pogruzit-ukrainu-vo-tmu-pered-nachalom-blekauta/
|
| 399 |
+
https://lenta.ru/news/2025/10/10/zelenskiy-rasskazal-o-nochnoy-atake-po-ukraine/
|
src/core/clean_core.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
src/core/core.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
src/core/preprocessed_core.jsonl
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
src/core/united_core.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
src/tokenizers_analysis.py
ADDED
|
@@ -0,0 +1,89 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import json
|
| 2 |
+
from nltk.tokenize import word_tokenize
|
| 3 |
+
from razdel import tokenize as razdel_tokenize
|
| 4 |
+
from snowballstemmer import RussianStemmer
|
| 5 |
+
import pymorphy2
|
| 6 |
+
import re
|
| 7 |
+
import time
|
| 8 |
+
from sentence_transformers import SentenceTransformer
|
| 9 |
+
from sklearn.metrics.pairwise import cosine_similarity
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
def tokenize_naive_space(text):
|
| 13 |
+
return text.split(' ')
|
| 14 |
+
|
| 15 |
+
|
| 16 |
+
def tokenize_regex(text):
|
| 17 |
+
return re.findall(r'[а-яА-ЯёЁ]+', text)
|
| 18 |
+
|
| 19 |
+
|
| 20 |
+
def tokenize_nltk(text):
|
| 21 |
+
return word_tokenize(text, language='russian')
|
| 22 |
+
|
| 23 |
+
|
| 24 |
+
def tokenize_razdel(text):
|
| 25 |
+
return list(map(lambda x: x.text, razdel_tokenize(text)))
|
| 26 |
+
|
| 27 |
+
|
| 28 |
+
def stem_snowball(tokens):
|
| 29 |
+
stemmer = RussianStemmer()
|
| 30 |
+
return stemmer.stemWords(tokens)
|
| 31 |
+
|
| 32 |
+
|
| 33 |
+
def lemmatize_pymorphy(words):
|
| 34 |
+
morph = pymorphy2.MorphAnalyzer()
|
| 35 |
+
lemmas = []
|
| 36 |
+
for word in words:
|
| 37 |
+
lemmas.append(morph.parse(word)[0].normal_form)
|
| 38 |
+
return lemmas
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
def calculate_oov(text, vocabulary):
|
| 42 |
+
words = text.split(' ')
|
| 43 |
+
oov_count = 0
|
| 44 |
+
for word in words:
|
| 45 |
+
if word not in vocabulary:
|
| 46 |
+
oov_count += 1
|
| 47 |
+
return oov_count / len(words)
|
| 48 |
+
|
| 49 |
+
|
| 50 |
+
def calculate_similarity(text1, text2, model):
|
| 51 |
+
embedding1 = model.encode(text1, convert_to_tensor=False).reshape(1, -1)
|
| 52 |
+
embedding2 = model.encode(text2, convert_to_tensor=False).reshape(1, -1)
|
| 53 |
+
return cosine_similarity(embedding1, embedding2)[0][0]
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
if __name__ == '__main__':
|
| 57 |
+
texts = []
|
| 58 |
+
with open('core/preprocessed_core.jsonl', encoding='utf-8') as file:
|
| 59 |
+
for line in file:
|
| 60 |
+
row = json.loads(line)
|
| 61 |
+
texts.append(row['text'])
|
| 62 |
+
n_articles = len(texts)
|
| 63 |
+
tokenizers = [tokenize_naive_space, tokenize_regex, tokenize_nltk, tokenize_razdel]
|
| 64 |
+
|
| 65 |
+
methods = []
|
| 66 |
+
for tokenizer in tokenizers:
|
| 67 |
+
methods.extend([[tokenizer], [tokenizer, stem_snowball], [tokenizer, lemmatize_pymorphy]])
|
| 68 |
+
|
| 69 |
+
csv_string = 'method;vocabulary volume;OOV percentage;processing speed;semantic consistency\n'
|
| 70 |
+
sim_model = SentenceTransformer('all-MiniLM-L6-v2')
|
| 71 |
+
for method in methods:
|
| 72 |
+
print('running', " + ".join(map(lambda x: x.__name__, method)))
|
| 73 |
+
start_time = time.time()
|
| 74 |
+
vocabulary = set()
|
| 75 |
+
similarities = []
|
| 76 |
+
for text in texts:
|
| 77 |
+
tokens = text
|
| 78 |
+
for func in method:
|
| 79 |
+
tokens = func(tokens)
|
| 80 |
+
similarities.append(calculate_similarity(text, ' '.join(tokens), sim_model))
|
| 81 |
+
with open(f'tokenized_texts/{"_".join(map(lambda x: x.__name__, method))}', 'a', encoding='utf-8') as file:
|
| 82 |
+
file.write(' '.join(tokens) + '\n')
|
| 83 |
+
vocabulary = vocabulary.union(tokens)
|
| 84 |
+
end_time = time.time()
|
| 85 |
+
csv_string += f'{" + ".join(map(lambda x: x.__name__, method))};{len(vocabulary)};'
|
| 86 |
+
csv_string += f'{calculate_oov(" ".join(texts), vocabulary)};{end_time - start_time};'
|
| 87 |
+
csv_string += f'{sum(similarities)/len(similarities)}\n'
|
| 88 |
+
with open('reports/tokenization_metrics.csv', 'w') as file:
|
| 89 |
+
file.write(csv_string)
|
src/tokenizers_trainer.py
ADDED
|
@@ -0,0 +1,239 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import re
|
| 3 |
+
import json
|
| 4 |
+
|
| 5 |
+
CORPUS_FILE = 'core/united_core.txt'
|
| 6 |
+
VOCAB_SIZE = 10000
|
| 7 |
+
OUTPUT_DIR = 'tokenizers'
|
| 8 |
+
os.makedirs(OUTPUT_DIR, exist_ok=True)
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
def simple_tokenize(text):
|
| 12 |
+
return re.findall(r'\S+', text)
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
def train_bpe(vocab_size, min_freq, corpus_path=None):
|
| 16 |
+
from tokenizers import Tokenizer
|
| 17 |
+
from tokenizers.models import BPE
|
| 18 |
+
from tokenizers.trainers import BpeTrainer
|
| 19 |
+
from tokenizers.pre_tokenizers import Whitespace
|
| 20 |
+
|
| 21 |
+
tokenizer = Tokenizer(BPE(unk_token='<UNK>'))
|
| 22 |
+
tokenizer.pre_tokenizer = Whitespace()
|
| 23 |
+
trainer = BpeTrainer(
|
| 24 |
+
vocab_size=vocab_size,
|
| 25 |
+
min_frequency=min_freq,
|
| 26 |
+
special_tokens=['<UNK>', '<NUM>', '<URL>', '<EMAIL>'],
|
| 27 |
+
continuing_subword_prefix='',
|
| 28 |
+
)
|
| 29 |
+
tokenizer.train(files=[corpus_path if corpus_path else CORPUS_FILE], trainer=trainer)
|
| 30 |
+
dir_path = f'{OUTPUT_DIR}/bpe_v{vocab_size//1000}k_f{min_freq}'
|
| 31 |
+
os.makedirs(dir_path, exist_ok=True)
|
| 32 |
+
tokenizer.save(os.path.join(dir_path, 'tokenizer.json'))
|
| 33 |
+
tokenizer_config = {
|
| 34 |
+
"added_tokens_decoder": {},
|
| 35 |
+
"unk_token": "<UNK>",
|
| 36 |
+
"cls_token": None,
|
| 37 |
+
"sep_token": None,
|
| 38 |
+
"mask_token": None,
|
| 39 |
+
"model_max_length": 512,
|
| 40 |
+
}
|
| 41 |
+
for token in ['<UNK>', '<NUM>', '<URL>', '<EMAIL>']:
|
| 42 |
+
t_id = str(tokenizer.encode(token).ids[0])
|
| 43 |
+
tokenizer_config['added_tokens_decoder'][t_id] = {
|
| 44 |
+
"content": token,
|
| 45 |
+
"lstrip": False,
|
| 46 |
+
"normalized": False,
|
| 47 |
+
"rstrip": False,
|
| 48 |
+
"single_word": False,
|
| 49 |
+
"special": True
|
| 50 |
+
}
|
| 51 |
+
with open(os.path.join(dir_path, "tokenizer_config.json"), "w", encoding="utf-8") as file:
|
| 52 |
+
json.dump(tokenizer_config, file, indent=2)
|
| 53 |
+
|
| 54 |
+
special_tokens_map = {
|
| 55 |
+
"unk_token": "<UNK>",
|
| 56 |
+
}
|
| 57 |
+
with open(os.path.join(dir_path, "special_tokens_map.json"), "w", encoding="utf-8") as file:
|
| 58 |
+
json.dump(special_tokens_map, file, indent=2)
|
| 59 |
+
|
| 60 |
+
return tokenizer
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
def train_wordpiece(vocab_size, min_freq, corpus_path=None):
|
| 64 |
+
from tokenizers import Tokenizer
|
| 65 |
+
from tokenizers.models import WordPiece
|
| 66 |
+
from tokenizers.trainers import WordPieceTrainer
|
| 67 |
+
from tokenizers.pre_tokenizers import Whitespace
|
| 68 |
+
|
| 69 |
+
tokenizer = Tokenizer(WordPiece(unk_token='<UNK>'))
|
| 70 |
+
tokenizer.pre_tokenizer = Whitespace()
|
| 71 |
+
trainer = WordPieceTrainer(
|
| 72 |
+
vocab_size=vocab_size,
|
| 73 |
+
min_frequency=min_freq,
|
| 74 |
+
special_tokens=['<UNK>', '<NUM>', '<URL>', '<EMAIL>'],
|
| 75 |
+
continuing_subword_prefix='',
|
| 76 |
+
)
|
| 77 |
+
tokenizer.train(files=[corpus_path if corpus_path else CORPUS_FILE], trainer=trainer)
|
| 78 |
+
path = f'{OUTPUT_DIR}/wordpiece_v{vocab_size}_f{min_freq}.json'
|
| 79 |
+
tokenizer.save(path)
|
| 80 |
+
return tokenizer
|
| 81 |
+
|
| 82 |
+
|
| 83 |
+
def train_unigram(vocab_size, min_freq, corpus_path=None):
|
| 84 |
+
import sentencepiece as spm
|
| 85 |
+
|
| 86 |
+
model_prefix = f'{OUTPUT_DIR}/unigram_v{vocab_size}_f{min_freq}'
|
| 87 |
+
try:
|
| 88 |
+
spm.SentencePieceTrainer.train(
|
| 89 |
+
input=corpus_path if corpus_path else CORPUS_FILE,
|
| 90 |
+
model_prefix=model_prefix,
|
| 91 |
+
vocab_size=vocab_size,
|
| 92 |
+
model_type='unigram',
|
| 93 |
+
character_coverage=0.9995,
|
| 94 |
+
pad_id=0, unk_id=1, bos_id=-1, eos_id=-1,
|
| 95 |
+
user_defined_symbols='<NUM>,<URL>,<EMAIL>',
|
| 96 |
+
shuffle_input_sentence=True,
|
| 97 |
+
input_sentence_size=100000,
|
| 98 |
+
normalization_rule_name='nmt_nfkc',
|
| 99 |
+
num_threads=8
|
| 100 |
+
)
|
| 101 |
+
sp = spm.SentencePieceProcessor()
|
| 102 |
+
sp.load(f'{model_prefix}.model')
|
| 103 |
+
return sp
|
| 104 |
+
except RuntimeError as e:
|
| 105 |
+
raise e
|
| 106 |
+
|
| 107 |
+
|
| 108 |
+
def fragmentation_rate(tokenizer_func, texts):
|
| 109 |
+
total_words = 0
|
| 110 |
+
fragmented = 0
|
| 111 |
+
for text in texts:
|
| 112 |
+
words = simple_tokenize(text)
|
| 113 |
+
for word in words:
|
| 114 |
+
tokens = tokenizer_func(word)
|
| 115 |
+
total_words += 1
|
| 116 |
+
if len(tokens) > 1:
|
| 117 |
+
fragmented += 1
|
| 118 |
+
return fragmented / total_words if total_words else 0
|
| 119 |
+
|
| 120 |
+
|
| 121 |
+
def compression_ratio(tokenizer_func, texts):
|
| 122 |
+
total_syms = 0
|
| 123 |
+
total_tokens = 0
|
| 124 |
+
for text in texts:
|
| 125 |
+
tokens = tokenizer_func(text)
|
| 126 |
+
total_syms += len(text)
|
| 127 |
+
total_tokens += len(tokens)
|
| 128 |
+
return total_syms / total_tokens if total_tokens else 0
|
| 129 |
+
|
| 130 |
+
|
| 131 |
+
def reconstruction_accuracy(tokenizer_obj, texts, model_type='hf'):
|
| 132 |
+
reconstructed_ok = 0
|
| 133 |
+
total_words = 0
|
| 134 |
+
|
| 135 |
+
for text in texts:
|
| 136 |
+
words = simple_tokenize(text)
|
| 137 |
+
for word in words:
|
| 138 |
+
total_words += 1
|
| 139 |
+
try:
|
| 140 |
+
if model_type == 'hf':
|
| 141 |
+
tokens = tokenizer_obj.encode(word).tokens
|
| 142 |
+
decoded = tokenizer_obj.decode(
|
| 143 |
+
tokenizer_obj.encode(word).ids,
|
| 144 |
+
skip_special_tokens=True
|
| 145 |
+
)
|
| 146 |
+
cleaned_decoded = re.sub(r'\s+', '', decoded.lower())
|
| 147 |
+
cleaned_word = re.sub(r'\s+|[^\w]', '', word.lower())
|
| 148 |
+
if cleaned_decoded == cleaned_word:
|
| 149 |
+
reconstructed_ok += 1
|
| 150 |
+
else:
|
| 151 |
+
pass
|
| 152 |
+
|
| 153 |
+
elif model_type == 'sp':
|
| 154 |
+
pieces = tokenizer_obj.encode_as_pieces(word)
|
| 155 |
+
decoded = ''.join(pieces).replace('▁', '')
|
| 156 |
+
cleaned_decoded = re.sub(r'\s+|[^\w]', '', decoded.lower())
|
| 157 |
+
cleaned_word = re.sub(r'\s+|[^\w]', '', word.lower())
|
| 158 |
+
if cleaned_decoded == cleaned_word:
|
| 159 |
+
reconstructed_ok += 1
|
| 160 |
+
|
| 161 |
+
except:
|
| 162 |
+
pass
|
| 163 |
+
|
| 164 |
+
return reconstructed_ok / total_words if total_words else 0
|
| 165 |
+
|
| 166 |
+
|
| 167 |
+
if __name__ == '__main__':
|
| 168 |
+
|
| 169 |
+
texts = []
|
| 170 |
+
with open('core/united_core.txt', encoding='utf-8') as file:
|
| 171 |
+
texts = file.readlines()
|
| 172 |
+
|
| 173 |
+
vocab_sizes = [8000, 16000, 32000]
|
| 174 |
+
min_freqs = [2, 3, 4, 5]
|
| 175 |
+
|
| 176 |
+
results = []
|
| 177 |
+
|
| 178 |
+
for vocab_size in vocab_sizes:
|
| 179 |
+
for min_freq in min_freqs:
|
| 180 |
+
print(f'vocab_size {vocab_size} min_freq {min_freq}')
|
| 181 |
+
|
| 182 |
+
try:
|
| 183 |
+
bpe = train_bpe(vocab_size, min_freq)
|
| 184 |
+
bpe_func = lambda x: bpe.encode(x).tokens
|
| 185 |
+
bpe_frag = fragmentation_rate(bpe_func, texts)
|
| 186 |
+
bpe_comp = compression_ratio(bpe_func, texts)
|
| 187 |
+
bpe_recon = reconstruction_accuracy(bpe, texts, model_type='hf')
|
| 188 |
+
results.append({
|
| 189 |
+
'model': 'BPE',
|
| 190 |
+
'vocab_size': vocab_size,
|
| 191 |
+
'min_freq': min_freq,
|
| 192 |
+
'fragmentation_rate': bpe_frag,
|
| 193 |
+
'compression_ratio': bpe_comp,
|
| 194 |
+
'reconstruction_acc': bpe_recon
|
| 195 |
+
})
|
| 196 |
+
except Exception as e:
|
| 197 |
+
print(f'BPE error: {e}')
|
| 198 |
+
|
| 199 |
+
try:
|
| 200 |
+
wp = train_wordpiece(vocab_size, min_freq)
|
| 201 |
+
wp_func = lambda x: wp.encode(x).tokens
|
| 202 |
+
wp_frag = fragmentation_rate(wp_func, texts)
|
| 203 |
+
wp_comp = compression_ratio(wp_func, texts)
|
| 204 |
+
wp_recon = reconstruction_accuracy(wp, texts, model_type='hf')
|
| 205 |
+
results.append({
|
| 206 |
+
'model': 'WordPiece',
|
| 207 |
+
'vocab_size': vocab_size,
|
| 208 |
+
'min_freq': min_freq,
|
| 209 |
+
'fragmentation_rate': wp_frag,
|
| 210 |
+
'compression_ratio': wp_comp,
|
| 211 |
+
'reconstruction_acc': wp_recon
|
| 212 |
+
})
|
| 213 |
+
except Exception as e:
|
| 214 |
+
print(f'WordPiece error: {e}')
|
| 215 |
+
|
| 216 |
+
try:
|
| 217 |
+
unigram = train_unigram(vocab_size, min_freq)
|
| 218 |
+
if unigram is not None:
|
| 219 |
+
uni_func = lambda x: unigram.encode_as_pieces(x)
|
| 220 |
+
uni_frag = fragmentation_rate(uni_func, texts)
|
| 221 |
+
uni_comp = compression_ratio(uni_func, texts)
|
| 222 |
+
uni_recon = reconstruction_accuracy(unigram, texts, model_type='sp')
|
| 223 |
+
|
| 224 |
+
results.append({
|
| 225 |
+
'model': 'Unigram',
|
| 226 |
+
'vocab_size': vocab_size,
|
| 227 |
+
'min_freq': min_freq,
|
| 228 |
+
'fragmentation_rate': uni_frag,
|
| 229 |
+
'compression_ratio': uni_comp,
|
| 230 |
+
'reconstruction_acc': uni_recon
|
| 231 |
+
})
|
| 232 |
+
except Exception as e:
|
| 233 |
+
print(f'Unigram error: {e}')
|
| 234 |
+
|
| 235 |
+
with open('reports/hf_sp_metrics.csv', 'w') as file:
|
| 236 |
+
file.write('model;vocab_size;min_freq;fragmentation_rate;compression_ratio;reconstruction_accuracy\n')
|
| 237 |
+
for r in results:
|
| 238 |
+
file.write(f'{r["model"]};{r["vocab_size"]};{r["min_freq"]};{round(r["fragmentation_rate"], 3)};')
|
| 239 |
+
file.write(f'{round(r["compression_ratio"], 3)};{round(r["reconstruction_acc"], 3)}\n')
|