Back to catalog

Tweety Tatar 7B

An open 7-billion-parameter language model with a native tokenizer for Tatar, part of the Tweeties research series for low-resource languages.

About this project

Created by an international multilingual-NLP research group
Adapted via trans-tokenization and fine-tuned on a Tatar corpus
The series also includes specialized model variants

Related projects

All projects in Language project

Aigiz Kunafin

AuthorLanguage project#Developer#Language#Technology#Text to speech#Language modelBashkirsTatars

Published 19 models and 13 datasets on Hugging Face for Bashkir and Tatar speech
Author of the tools bashspell, bashkort_translate_bot and bashkir-stress
Winner of the Enterprise RAG Challenge

Fastmorph

Language project#Language#Technology#Language model#Open sourceTatars

Created for the Written Tatar Language Corpus
Searches by word forms, lemmas, grammatical tags, and patterns
Supports distance constraints between words
Source code published under the GPL-3.0 license

Written Corpus of the Tatar Language

Language project#Language#Technology#Language modelTatars

The corpus contains hundreds of millions of Tatar words
Supports search by word form and grammatical features
Includes a thesaurus and a spell-checking service

Муэллиф