Dil modeli
90,54%
Gemma 3 27B modelinden +4,47 öňde
- 01Airun LLM90,54
- 02Gemma 3 27B86,07
- 03DeepSeek-R1 32B74,99
- 04Qwen3 30B69,56
Jogaplaryň takyklygy, %. Ýumuşlar gyrgyz dilinde.
Bäş ölçeg — bäş birinji orun.
DeňeşdirilenlerGemma 3 27BDeepSeek-R1 32BQwen3 30BKani TTSAitil.kgClaude Opus 4.6Aitil.kg TranslateGPT-5.4Grok 4.20
90,54%
Gemma 3 27B modelinden +4,47 öňde
Jogaplaryň takyklygy, %. Ýumuşlar gyrgyz dilinde.
6,39%
Iň ýakyn bäsdeşden 3,2 esse az ýalňyş
WER — her 100 sözdäki ýalňyşlar. Diňe kiril.
52,9
Claude Opus 4.6 modelinden +6,9 öňde
BLEU — nusga terjimä ýakynlyk.
Emeli aňyň gyrgyz we Merkezi Aziýanyň beýleki dillerine düşünişini öwrenýäris.
MeýdançalarMeLLMMRLLoResMTClimateNLParXiv
MeLLM 2026·ACL Anthology2026.mellm-1.5
Zarina Uvalieva, Bektemir Kumarbai Uulu, Adilet Metinov, Tynchtykbek Tashbaltaev, Nurtilek Alibekov
Abstract
Text normalization, the task of converting noisy, informal text into a standardized form - is a fundamental preprocessing step for many NLP applications. Despite the growing need for Kyrgyz language processing tools, to the best of our knowledge, no prior work has addressed automatic text normalization for Kyrgyz, a morphologically rich, low-resource Turkic language. In this paper, we present the first systematic study of Kyrgyz text normalization. We collect a dataset of 1.67 million noisy–clean text pairs sourced from YouTube comments, Instagram posts, and Telegram channels, where users frequently write without punctuation, capitalization, or standard spelling. Pairs were annotated with Gemini 3 Pro; the 1,000-example test set was fully verified by two native Kyrgyz speakers with adjudication, and a random subset of the training data was spot-checked, while the full 1.67M training set was not verified exhaustively. For continual pre-training, we additionally use a 538 MB Kyrgyz corpus compiled from news portals and books. We evaluate five systems: a rule-based baseline, zero-shot mT5, a fine-tuned mT5-small model, a continually pre-trained mT5-small followed by fine-tuning, and zero-shot Gemma 4. Our experiments show that fine-tuned mT5-small achieves a CER of 0.0796, outperforming the rule-based baseline (CER 0.2029), zero-shot mT5 (CER 0.9887), and zero-shot Gemma 4 (CER 0.1620), a roughly 32× larger model in a fine-tuned vs. zero-shot setting. Human evaluation by two native Kyrgyz speakers confirms these results, with fine-tuned mT5-small rated as correct in 99.8% of cases. We further analyze why continual pre-training with span corruption does not improve over direct fine-tuning, finding hallucination in 35/40 of the inspected failure cases (87.5%, 95% Wilson CI [74%, 95%]).
Proceedings of the 1st Workshop on Multilinguality in the Era of Large Language Models (MeLLM 2026), pages 52–62July 2026 ©2026 Association for Computational Linguistics52
Sosial ulgamlardan 1,67 mln jübüt boýunça ilkinji ulgamlaýyn barlag: neýron tor kadalardan 2,5 esse az ýalňyşýar.
Ýol, söwda, resminamalar we kanunlar: şäherde gerek zatlaryň hemmesi.
Prodiňe Airun Pro abunalygy bilen
Tutuş ýurt Airun-a birikdirildi
SanawdaBanklar 2Döwlet 3Bilim 8Biznes 9
Banklar
Bakai Bank
Banklar
Берекет Банк
Döwlet
Жогорку Кенеш
Döwlet
Министерство юстиции
Döwlet
Мэрия Бишкека
Bilim
Университет Манас
Bilim
Университет Ала-Тоо
Bilim
КГМА
Bilim
КРСУ
Bilim
МУК
Bilim
AUCA
Bilim
Compass College
Bilim
Nomad College
Biznes
mTicket.kg
Biznes
Etnomedia
Biznes
1.1 studio
Biznes
OnClinic
Biznes
ArtWin
Biznes
Devolution
Biznes
Profile.kg
Biznes
Bakai Store
Biznes
Bakai Travel
Biz barada ýazanlar
The Times of Central AsiaSputnik КыргызстанToday.kgEconomist.kgBanks.kgReporter.kgKaktus.mediaTHE TECHАкчабарРамблер/новости
Özüňiz üçin: brauzerde we telefonda. Önüm üçin: API arkaly Airun modelleri.