Russian-Bashkir Machine Translation

Your task is to translate from Russian to Bashkir language.

• Test file is formatted as follows: id,question. You can find it here.
• Your submission should be formatted as follows: id,submission. You should place you translated text in the submission field. Sample submission can be found here. To make sample submission we used Google Gemini.

You can use Bashkir-Russian Parallel Corpus (1,200,000 sentence pairs) as training one. Link: https://huggingface.co/datasets/AigizK/bashkir-russian-parallel-corpora

Sources:

• Fiction and school textbooks verified via Telegram bot by native speakers
• Bashkir Encyclopedia articles
• News from bashinform.ru
• Mixed domain content (TMX corpus, literature)

Purpose: Training baseline translation models

License: CC-BY-4.0

Existing fine-tuned models:

• NLLB-200
• https://huggingface.co/slone/bert-base-multilingual-cased-bak-rus-similarity - This is a text pair classifier, trained to predict whether a Bashkir sentence and a Russian sentence have the same meaning
• https://huggingface.co/slone/LaBSE-shallow-distilled-bak - compute sentence embedings compatible with LaBSE and to do masked language modelling