The code for training mCOLT/mRASP2, a multilingual neural machine translation training method, implemented based on fairseq.

mRASP2: paper blog

mRASP: paper, code

We have released two versions, this version is the original one. In this implementation:

• You should first merge all data, by pre-pending language token before each sentence to indicate the language.
• AA/RAS muse be done off-line (before binarize), check this toolkit.

New implementation: https://github.com/PANXiao1994/mRASP2/tree/new_impl

• Acknowledgement: This work is supported by Bytedance. We thank Chengqi for uploading all files and checkpoints.

mRASP2/mCOLT, representing multilingual Contrastive Learning for Transformer, is a multilingual neural machine translation model that supports complete many-to-many multilingual machine translation. It employs both parallel corpora and multilingual corpora in a unified training framework. For detailed information please refer to the paper.

pip install -r requirements.txt
# install fairseq
git clone https://github.com/pytorch/fairseq
cd fairseq
pip install --editable ./

We release our preprocessed training data and checkpoints in the following.

We merge 32 English-centric language pairs, resulting in 64 directed translation pairs in total. The original 32 language pairs corpus contains about 197M pairs of sentences. We get about 262M pairs of sentences after applying RAS, since we keep both the original sentences and the substituted sentences. We release both the original dataset and dataset after applying RAS.

• Please note that the provided checkpoint is sightly different from that in the paper. In the following sections, we report the results of the provided checkpoints.

We report tokenized BLEU in the following table. Please click the model links to download. It is in pytorch format. (check eval.sh for details)

We report tokenized BLEU in the following table. (check eval.sh for details)

• row: source language
• column: target language We report sacreBLEU in the following table.

export NUM_GPU=4 && bash train_w_mono.sh ${model_config}

• We give example of ${model_config} in ${PROJECT_REPO}/examples/configs/parallel_mono_12e12d_contrastive.yml

• You must pre-pend the corresponding language token to the source side before binarize the test data.

fairseq-generate ${test_path} \
    --user-dir ${repo_dir}/mcolt \
    -s ${src} \
    -t ${tgt} \
    --skip-invalid-size-inputs-valid-test \
    --path ${ckpts} \
    --max-tokens ${batch_size} \
    --task translation_w_langtok \
    ${options} \
    --lang-prefix-tok "LANG_TOK_"`echo "${tgt} " | tr '[a-z]' '[A-Z]'` \
    --max-source-positions ${max_source_positions} \
    --max-target-positions ${max_target_positions} \
    --nbest 1 | grep -E '[S|H|P|T]-[0-9]+' > ${final_res_file}
python3 ${repo_dir}/scripts/utils.py ${res_file} ${ref_file} || exit 1;

We use the bilingual synonym dictionaries provised by MUSE.

We generate multilingual synonym dictionaries using this script, and apply RAS using this script.

Please contact me via e-mail [email protected] or via wechat/zhihu or join the slack group!

Please cite as:

@inproceedings{mrasp2,
  title = {Contrastive Learning for Many-to-many Multilingual Neural Machine Translation},
  author= {Xiao Pan and
           Mingxuan Wang and
           Liwei Wu and
           Lei Li},
  booktitle = {Proceedings of ACL 2021},
  year = {2021},
}