Quick Installation

We recommend python 3.6 to run COMET.

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

Simple installation from PyPI

pip install unbabel-comet

To develop locally:

git clone https://github.com/Unbabel/COMET
pip install -r requirements.txt
pip install -e .

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 -h hyp.en -r ref.en

You can export your results to a JSON file using the --to_json flag and select another model/metric with --model.

comet score -s src.de -h hyp.en -r ref.en --model wmt-large-hter-estimator --to_json segments.json

Via Python:

from comet.models import download_model
model = download_model("wmt-large-da-estimator-1719")
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"
    }
]
model.predict(data, cuda=True, show_progress=True)

Scoring MT outputs with MCD runs

To run COMET with multiple MCD runs:

 #!/bin/bash
 
GPU_N=3

SCORES=/path/to/your/result/folder
DATA=/path/to/your/data/folder

N=100
D=0.1
N_REFS=1

SRC=src.txt
MT=mt.txt
REF=ref.txt

MODEL=wmt-large-da-estimator-1719

echo Starting the process...

CUDA_VISIBLE_DEVICES=$GPU_N comet score \
  -s $DATA/sources/$SRC \
  -h $DATA/system-outputs/$MT \
  -r $DATA/references/$REF \
  --to_json $SCORES/filename.json \
  --n_refs $N_REFS \
  --n_dp_runs $N \
  --d_enc $D \
  --d_pool $D \
  --d_ff1 $D \
  --d_ff2 $D \
  --model $MODEL 

This will run the model with a set of hyperparameters defined above. Here is the description of the main scoring arguments:

-s: Source segments.
-h: MT outputs.
-r: Reference segments.
--to_json: Creates and exports model predictions to a JSON file.
--n_refs: default=1. Number of references used during inference.
--n_dp_runs: default=30. Number of dropout runs at test time.
--d_enc: default=0.1. Dropout value for the encoder.
--d_pool: default=0.1. Dropout value for the layerwise pooling layer.
--d_ff1: default=0.1. Dropout value for the 1st feed forward layer.
--d_ff2: default=0.1. Dropout value for the 2nd feed forward layer.
--model: Name of the pretrained model OR path to a model checkpoint.

To know more about the rest of the parameters and their default values, take a look at the comet/cli.py file.

How to Reproduce and Evaluate Experiments

MCD and DEE

pointers for csores files

Multi-reference

link to prism github link to translations

Precision/Recall

pointer to jupyter notebook

Model Zoo:

Model	Description
↑wmt-large-da-estimator-1719	RECOMMENDED: Estimator model build on top of XLM-R (large) trained on DA from WMT17, WMT18 and WMT19
↑wmt-base-da-estimator-1719	Estimator model build on top of XLM-R (base) trained on DA from WMT17, WMT18 and WMT19
↓wmt-large-hter-estimator	Estimator model build on top of XLM-R (large) trained to regress on HTER.
↓wmt-base-hter-estimator	Estimator model build on top of XLM-R (base) trained to regress on HTER.
↑emnlp-base-da-ranker	Translation ranking model that uses XLM-R to encode sentences. This model was trained with WMT17 and WMT18 Direct Assessments Relative Ranks (DARR).

QE-as-a-metric:

Model	Description
wmt-large-qe-estimator-1719	Quality Estimator model build on top of XLM-R (large) trained on DA from WMT17, WMT18 and WMT19.

Train your own Metric:

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

comet train -f {config_file_path}.yaml

Supported encoders:

• Learning Joint Multilingual Sentence Representations with Neural Machine Translation
• Massively Multilingual Sentence Embeddings for Zero-Shot Cross-Lingual Transfer and Beyond
• BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
• XLM-R: Unsupervised Cross-lingual Representation Learning at Scale

Tensorboard:

Launch tensorboard with:

tensorboard --logdir="experiments/"

Download Command:

To download public available corpora to train your new models you can use the download command. For example to download the APEQUEST HTER corpus just run the following command:

comet download -d apequest --saving_path data/

Publications

@inproceedings{rei-etal-2020-comet,
    title = "{COMET}: A Neural Framework for {MT} Evaluation",
    author = "Rei, Ricardo  and
      Stewart, Craig  and
      Farinha, Ana C  and
      Lavie, Alon",
    booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)",
    month = nov,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.emnlp-main.213",
    pages = "2685--2702",
}

@inproceedings{rei-EtAl:2020:WMT,
  author    = {Rei, Ricardo  and  Stewart, Craig  and  Farinha, Ana C  and  Lavie, Alon},
  title     = {Unbabel's Participation in the WMT20 Metrics Shared Task},
  booktitle      = {Proceedings of the Fifth Conference on Machine Translation},
  month          = {November},
  year           = {2020},
  address        = {Online},
  publisher      = {Association for Computational Linguistics},
  pages     = {909--918},
}

@inproceedings{stewart-etal-2020-comet,
    title = "{COMET} - Deploying a New State-of-the-art {MT} Evaluation Metric in Production",
    author = "Stewart, Craig  and
      Rei, Ricardo  and
      Farinha, Catarina  and
      Lavie, Alon",
    booktitle = "Proceedings of the 14th Conference of the Association for Machine Translation in the Americas (Volume 2: User Track)",
    month = oct,
    year = "2020",
    address = "Virtual",
    publisher = "Association for Machine Translation in the Americas",
    url = "https://www.aclweb.org/anthology/2020.amta-user.4",
    pages = "78--109",
}