1 Star 0 Fork 0

明忌/delight

加入 Gitee
与超过 1200万 开发者一起发现、参与优秀开源项目,私有仓库也完全免费 :)
免费加入
克隆/下载
prepare_nmt_dataset.sh 1.23 KB
一键复制 编辑 原始数据 按行查看 历史
Sachin Mehta 提交于 2020-08-26 11:16 . Updated the data-bin location
#!/usr/bin/env bash
DATASET=$1
if [ "$DATASET" == "" ]; then
echo "Please specify the dataset name. Supported choices: wmt14_en_de, wmt14_en_fr"
exit
fi
##
## WMT14 En2De dataset
##
if [ "$DATASET" == "wmt14_en_de" ]; then
# Download and prepare the data
cd examples/translation/
bash prepare_wmt14_en2de.sh
cd ../..
# Preprocess/binarize the data
TEXT=examples/translation/wmt14_en_de
fairseq-preprocess \
--source-lang en --target-lang de \
--trainpref $TEXT/train --validpref $TEXT/valid --testpref $TEXT/test \
--destdir data-bin/wmt14_en_de --thresholdtgt 0 --thresholdsrc 0 \
--workers 20 --joined-dictionary
##
## WMT14 En2Fr dataset
##
elif [ "$DATASET" == "wmt14_en_fr" ]; then
cd examples/translation/
bash prepare_wmt14_en2fr.sh
cd ../..
# Binarize the dataset
TEXT=examples/translation/wmt14_en_fr
fairseq-preprocess \
--source-lang en --target-lang fr \
--trainpref $TEXT/train --validpref $TEXT/valid --testpref $TEXT/test \
--destdir data-bin/wmt14_en_fr --thresholdtgt 0 --thresholdsrc 0 \
--workers 60 --joined-dictionary
else
echo "Only these datasets are supported (wmt14_en_de, wmt14_en_fr)"
fi
马建仓 AI 助手
尝试更多
代码解读
代码找茬
代码优化
1
https://gitee.com/mingjizangmin/delight.git
git@gitee.com:mingjizangmin/delight.git
mingjizangmin
delight
delight
master

搜索帮助

23e8dbc6 1850385 7e0993f3 1850385