Hello!

My name is Vladislav Grigorian. Together with Abu-Viskhan Umishov we maintained google/wmt24pp preparation for the Chechen language.

As a reference for the translation we used Russian "target" version of the dataset.

For the translation of the dataset, we applied the same guidelines we used for Chechen smol and BOUQuET we translated earlier. Here they are down below in English:

1. The sentences must be translated accurately and sound natural in Chechen.
2. Consider the context of the other sentences in the paragraph when translating each sentence.
3. If you have difficulty with a sentence, skip it.
4. Use dictionaries if needed, but do not rely on Google Translate or any other machine translation systems.
5. Use 1 (Digit One) for Cyrillic Letter Palochka and Cyrillic Small Letter Palochka.
6. Maintain the capitalization of the original text. Example: "Кот - Цициг", "КОТ - ЦИЦИГ".
7. For Chechen words, use only Cyrillic characters (and 1 for both cases of Palochka). An example of what not to do: "парень - k1ант", here letter k is Latin.
8. Preserve the original untranslatable parts of the text, such as emojis, IDs and words that were not translated in the Russian reference, exactly.

After the translation was finished we substituted Digit One for Small and Capital Palochka letters where grammatical correctness required that. Please do not unify Cyrillic Palochka casing for Chechen as it is not grammatically correct for the language.

We also found out a small inconsistency in the Russian version of the benchmark: in "test-en-news_seattle_times.800119" documents both Cyrillic and Latin letters were used for plane model IDs - in the Chechen version we only used Latin letters for that.

It wasn't clear enough to us if we have to put the final variant of our translation into "original_target" or "target" column - so we duplicated the Chechen version for both.

Please direct any inquiries regarding this translation to real.vladislav.grigorian@gmail.com or abuviskhanumishov@gmail.com.