Merge pull request #755 from PaddlePaddle/ted

fix ted egs

examples/ted_en_zh/t0/.gitignore

@@ -0,0 +1,3 @@
+TED-En-Zh
+data
+exp

examples/ted_en_zh/t0/README.md

@@ -0,0 +1,10 @@
+
+# TED En-Zh
+
+## Dataset
+
+| Data Subset | Duration in Seconds |
+| --- | --- |
+| data/manifest.train | 0.942 ~ 60   |
+| data/manifest.dev   | 1.151 ~ 39   |  
+| data/manifest.test  | 1.1 ~ 42.746 |

examples/ted_en_zh/t0/local/data.sh

@@ -7,17 +7,18 @@ stop_stage=100
 nbpe=8000
 bpemode=unigram
 bpeprefix="data/bpe_${bpemode}_${nbpe}"
-DATA_DIR= 
+data_dir=/mnt/dataset/TED_EnZh
 
 
 source ${MAIN_ROOT}/utils/parse_options.sh
 
-
-mkdir -p data
 TARGET_DIR=${MAIN_ROOT}/examples/dataset
 mkdir -p ${TARGET_DIR}
+mkdir -p data
 
-if [ ! -d ${SOURCE_DIR} ]; then
+
+if [ ${stage} -le -1 ] && [ ${stop_stage} -ge -1 ]; then
+    if [ ! -e ${data_dir} ]; then
         echo "Error: Dataset is not avaiable. Please download and unzip the dataset"
         echo "Download Link: https://pan.baidu.com/s/18L-59wgeS96WkObISrytQQ Passwd: bva0"
         echo "The tree of the directory should be:"
@@ -33,11 +34,10 @@ if [ ! -d ${SOURCE_DIR} ]; then
         exit 1
     fi
 
-if [ ${stage} -le -1 ] && [ ${stop_stage} -ge -1 ]; then
     # generate manifests
     python3 ${TARGET_DIR}/ted_en_zh/ted_en_zh.py \
     --manifest_prefix="data/manifest" \
-    --src_dir="${DATA_DIR}" 
+    --src_dir="${data_dir}"
 
     echo "Complete raw data pre-process."
 fi

examples/ted_en_zh/t0/run.sh

@@ -16,7 +16,7 @@ echo "checkpoint name ${ckpt}"
 
 if [ ${stage} -le 0 ] && [ ${stop_stage} -ge 0 ]; then
     # prepare data
-    bash ./local/data.sh --DATA_DIR ${data_path} || exit -1
+    bash ./local/data.sh --data_dir ${data_path} || exit -1
 fi
 
 if [ ${stage} -le 1 ] && [ ${stop_stage} -ge 1 ]; then