Welcome to mirror list, hosted at ThFree Co, Russian Federation.

preprocess-de.sh « wmt16_systems « models - github.com/marian-nmt/marian-regression-tests.git - Unnamed repository; edit this file 'description' to name the repository.
summaryrefslogtreecommitdiff
blob: cd9139e24da393421ba4a352d9cdd6962fff05b4 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
#!/bin/bash

root="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
moses_scripts=$root/../../tools/moses-scripts
subword_nmt=$root/../../tools/subword-nmt

model_dir=$root/de-en

$moses_scripts/scripts/tokenizer/normalize-punctuation.perl -l de \
    | $moses_scripts/scripts/tokenizer/tokenizer.perl -l de -penn -threads 16 \
    | $moses_scripts/scripts/recaser/truecase.perl -model $model_dir/truecase-model.de \
    | $subword_nmt/apply_bpe.py -c $model_dir/deen.bpe