unbabel-comet 1.0.0rc2

High-quality Machine Translation Evaluation

Quick Installation

Detailed usage examples and instructions can be found in the Full Documentation.

Simple installation from PyPI

Pre-release of version 1.0:

pip install unbabel-comet==1.0.0rc2

To develop locally install Poetry and run the following commands:

git clone https://github.com/Unbabel/COMET
poetry install

Scoring MT outputs:

Via Bash:

Examples from WMT20:

echo -e "Dem Feuer konnte Einhalt geboten werden\nSchulen und Kindergärten wurden eröffnet." >> src.de
echo -e "The fire could be stopped\nSchools and kindergartens were open" >> hyp.en
echo -e "They were able to control the fire.\nSchools and kindergartens opened" >> ref.en

comet-score -s src.de -t hyp.en -r ref.en

You can select another model/metric with the --model flag and for reference-free (QE-as-a-metric) models you don't need to pass a reference.

comet-score -s src.de -t hyp.en -r ref.en --model wmt21-comet-qe-da

Following the work on Uncertainty-Aware MT Evaluation you can use the --mc_dropout flag to get a variance/uncertainty value for each segment score. If this value is high, it means that the metric as less confidence is that prediction.

comet-score -s src.de -t hyp.en -r ref.en --mc_dropout 100

Languages Covered:

All the above mentioned models are build on top of XLM-R which cover the following languages:

Afrikaans, Albanian, Amharic, Arabic, Armenian, Assamese, Azerbaijani, Basque, Belarusian, Bengali, Bengali Romanized, Bosnian, Breton, Bulgarian, Burmese, Burmese, Catalan, Chinese (Simplified), Chinese (Traditional), Croatian, Czech, Danish, Dutch, English, Esperanto, Estonian, Filipino, Finnish, French, Galician, Georgian, German, Greek, Gujarati, Hausa, Hebrew, Hindi, Hindi Romanized, Hungarian, Icelandic, Indonesian, Irish, Italian, Japanese, Javanese, Kannada, Kazakh, Khmer, Korean, Kurdish (Kurmanji), Kyrgyz, Lao, Latin, Latvian, Lithuanian, Macedonian, Malagasy, Malay, Malayalam, Marathi, Mongolian, Nepali, Norwegian, Oriya, Oromo, Pashto, Persian, Polish, Portuguese, Punjabi, Romanian, Russian, Sanskri, Scottish, Gaelic, Serbian, Sindhi, Sinhala, Slovak, Slovenian, Somali, Spanish, Sundanese, Swahili, Swedish, Tamil, Tamil Romanized, Telugu, Telugu Romanized, Thai, Turkish, Ukrainian, Urdu, Urdu Romanized, Uyghur, Uzbek, Vietnamese, Welsh, Western, Frisian, Xhosa, Yiddish.

Thus, results for language pairs containing uncovered languages are unreliable!

Scoring within Python:

COMET implements the Pytorch-Lightning model interface which means that you'll need to initialize a trainer in order to run inference.

import torch
from comet import download_model, load_from_checkpoint
from pytorch_lightning.trainer.trainer import Trainer
from torch.utils.data import DataLoader

model = load_from_checkpoint(
  download_model("wmt20-comet-da")
)
data = [
    {
        "src": "Dem Feuer konnte Einhalt geboten werden",
        "mt": "The fire could be stopped",
        "ref": "They were able to control the fire."
    },
    {
        "src": "Schulen und Kindergärten wurden eröffnet.",
        "mt": "Schools and kindergartens were open",
        "ref": "Schools and kindergartens opened"
    }
]
data = [dict(zip(data, t)) for t in zip(*data.values())]
dataloader = DataLoader(
  dataset=data,
  batch_size=16,
  collate_fn=lambda x: model.prepare_sample(x, inference=True),
  num_workers=4,
)
trainer = Trainer(gpus=1, deterministic=True, logger=False)
predictions = trainer.predict(
  model, dataloaders=dataloader, return_predictions=True
)
predictions = torch.cat(predictions, dim=0).tolist()

Note: Using the python interface you will get a list of segment-level scores. You can obtain the corpus-level score by averaging the segment-level scores

Model Zoo:

:TODO: Update model zoo after the shared task.

Model	Description
wmt20-comet-da	DEFAULT: Regression model build on top of XLM-R (large) trained on DA from WMT17, to WMT19. This model was presented at the WMT20 Metrics shared task: rei et al, 2020. Same as wmt-large-da-estimator-1719 from previous versions.
emnlp20-comet-rank	Translation Ranking model build on top of XLM-R (base) trained with DARR from WMT17 and WMT18. This model was presented at EMNLP20: rei et al, 2020.
wmt21-comet-da	Regression model build on top of XLM-R (large) trained on DA from WMT15, to WMT20. This model was presented at the WMT21 Metrics shared task.
wmt21-comet-mqm	Regression model build on top of XLM-R (large) trained to maximize correlation with MQM annotations from freitag et al, 2020.

QE-as-a-metric:

The following models can be used to assess translation quality without the need of references!

Model	Description
wmt21-comet-qe-da	Reference-free Regression model build on top of XLM-R (large) trained on DA from WMT15, to WMT20. This model was presented at the WMT21 Metrics shared task.
wmt21-comet-qe-mqm	Reference-free Regression model build on top of XLM-R (large) trained to maximize correlation with MQM annotations from freitag et al, 2020.

Lightweight models:

One of the remaining redeeming qualities of automated metrics such as BLEU is that they are incredibly lightweight. For that reason we have been developing COMETinho's, lightweight versions of the previous models.

Model	Description
wmt21-cometinho-da	Regression model build on top of XLM-R (large) trained on DA from WMT15, to WMT20. This model was presented at the WMT21 Metrics shared task.
wmt21-cometinho-mqm	Regression model build on top of XLM-R (large) trained to maximize correlation with MQM annotations from freitag et al, 2020.

Train your own Metric:

Instead of using pretrained models your can train your own model with the following command:

comet-train -cfg configs/models/{your_model_config}.yaml

Tensorboard:

Launch tensorboard with:

tensorboard --logdir="lightning_logs/"

unittest:

In order to run the toolkit tests you must run the following command:

coverage run --source=comet -m unittest discover
coverage report -m

Publications

• COMET: A Neural Framework for MT Evaluation
• Unbabel's Participation in the WMT20 Metrics Shared Task
• COMET - Deploying a New State-of-the-art MT Evaluation Metric in Production