Helsinki-NLP/opus-mt_tiny_eng-ell

OPUS-MT-tiny-eng-ell

Distilled model from a Tatoeba-MT Teacher:Tatoeba-MT-models/eng-ell/opusTCv20210807+bt_transformer-big_2022-03-13, which has been trained on the Tatoeba dataset.

We used the OpusDistillery to train new a new student with the tiny architecture, with a regular transformer decoder. For training data, we used Tatoeba. The configuration file fed into OpusDistillery can be found here.

How to run

from transformers import MarianMTModel, MarianTokenizer
model_name = "Helsinki-NLP/opus-mt_tiny_eng-ell"
tokenizer = MarianTokenizer.from_pretrained(model_name)
model = MarianMTModel.from_pretrained(model_name)
tok = tokenizer("The area is also home to species of animals and birds with a wide variety.", 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+	27.3	53.9	0.8809
Bouquet	43.4	65.7	0.9064

Student

testset	BLEU	chr-F	COMET
Flores+	25.2	52.2	0.8530
Bouquet	40.9	63.0	0.8784

Marian models

We also provide Marian-compatible versions of this model. To use them, compile Marian and run decoding with marian-decoder, for example:

marian-decoder \
  -i input.txt \
  -c final.model.npz.best-perplexity.npz.decoder.yml \
  -m final.model.npz.best-perplexity.npz \
  -v vocab.spm vocab.spm