---
license: apache-2.0
datasets:
- Helsinki-NLP/tatoeba
- openlanguagedata/flores_plus
- facebook/bouquet
language:
- en
- ru
metrics:
- bleu
- comet
- chrf
pipeline_tag: translation
---

# OPUS-MT-tiny-rus-eng

Distilled model from a Tatoeba-MT Teacher: [OPUS-MT-models/ru-en/opus-2020-02-26](https://object.pouta.csc.fi/OPUS-MT-models/ru-en/opus-2020-02-26.zip), which has been trained on the [Tatoeba](https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/data) dataset.

We used the [OpusDistillery](https://github.com/Helsinki-NLP/OpusDistillery) to train new a new student with the tiny architecture, with a regular transformer decoder.
For training data, we used [Tatoeba](https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/data).
The configuration file fed into OpusDistillery can be found [here](https://github.com/Helsinki-NLP/OpusDistillery/blob/main/configs/opustranslate_hf/config.op.ru-en.yml).

## How to run
```python
from transformers import MarianMTModel, MarianTokenizer
model_name = "Helsinki-NLP/opus-mt_tiny_rus-eng"
tokenizer = MarianTokenizer.from_pretrained(model_name)
model = MarianMTModel.from_pretrained(model_name)
tok = tokenizer("Это привело к тому, что два вида рыб вымерли, а два других, в том числе горбатый голавль, попали под угрозу исчезновения.", return_tensors="pt").input_ids
output = model.generate(tok)[0]
tokenizer.decode(output, skip_special_tokens=True)
```

## Benchmarks
### Teacher
| testset               | BLEU  | chr-F | COMET| 
|-----------------------|-------|-------|-------|
| Flores+ 	| 29.4 | 57.7| 0.8172 |
| Bouquet 	|33.4 	| 56.6 |0.8235 |

### Student

| testset               | BLEU  | chr-F | COMET |
|-----------------------|-------|-------|-------|
| Flores+ 	| 28.4 	| 56.8 | 0.8220 |
| Bouquet 	|31.1|54.1| 0.8151 |