fairseq=/path/to/fairseq
cd $fairseq

# Download 615M param model.
wget https://dl.fbaipublicfiles.com/flores101/pretrained_models/flores101_mm100_615M.tar.gz

# Extract 
tar -xvzf flores101_mm100_615M.tar.gz

fairseq=/path/to/fairseq
cd $fairseq

# Download example dataset From German to French
sacrebleu --echo src -l de-fr -t wmt19 | head -n 20 > raw_input.de-fr.de
sacrebleu --echo ref -l de-fr -t wmt19 | head -n 20 > raw_input.de-fr.fr

for lang in de fr ; do
    python scripts/spm_encode.py \
        --model flores101_mm100_615M/sentencepiece.bpe.model \
        --output_format=piece \
        --inputs=raw_input.de-fr.${lang} \
        --outputs=spm.de-fr.${lang}
done

fairseq-preprocess \
    --source-lang de --target-lang fr \
    --testpref spm.de-fr \
    --thresholdsrc 0 --thresholdtgt 0 \
    --destdir data_bin \
    --srcdict flores101_mm100_615M/dict.txt --tgtdict flores101_mm100_615M/dict.txt

fairseq-generate \
    data_bin \
    --batch-size 1 \
    --path flores101_mm100_615M/model.pt \
    --fixed-dictionary flores101_mm100_615M/dict.txt \
    -s de -t fr \
    --remove-bpe 'sentencepiece' \
    --beam 5 \
    --task translation_multi_simple_epoch \
    --lang-pairs flores101_mm100_615M/language_pairs.txt \
    --decoder-langtok --encoder-langtok src \
    --gen-subset test \
    --fp16 \
    --dataset-impl mmap \
    --distributed-world-size 1 --distributed-no-spawn