Views
No views yet
| Current Model | Architecture | Focus |
|---|---|---|
| 🔴 Large Model | NLLB-1.3B (QLoRA) | Best Quality (SOTA) |
| 🟡 Medium Model | M2M-100 (418M) | Balanced |
| 🟢 Small (This Model) | MarianMT (Full FT) | Fastest / CPU Friendly |
>>bak<< to the source sentence.1from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
2
3model_id = "Voldis/marian-rus-bak"
4tokenizer = AutoTokenizer.from_pretrained(model_id)
5model = AutoModelForSeq2SeqLM.from_pretrained(model_id)
6
7# Note the prefix!
8text = ">>bak<< Добрый день, друзья!"
9inputs = tokenizer(text, return_tensors="pt")
10
11outputs = model.generate(**inputs, max_length=128)
12print(tokenizer.batch_decode(outputs, skip_special_tokens=True)[0])1@inproceedings{tyurin-2026-devlake,
2 title = "{D}ev{L}ake at {L}o{R}es{MT} 2026: The Impact of Pre-training and Model Scale on {R}ussian-{B}ashkir Low-Resource Translation",
3 author = "Tyurin, Vyacheslav",
4 booktitle = "Proceedings for the Ninth Workshop on Technologies for Machine Translation of Low Resource Languages (LoResMT 2026)",
5 month = mar,
6 year = "2026",
7 address = "Rabat, Morocco",
8 publisher = "Association for Computational Linguistics",
9 url = "https://aclanthology.org/2026.loresmt-1.18",
10 doi = "10.18653/v1/2026.loresmt-1.18",
11 pages = "209--212",
12}