diff --git a/download.sh b/download.sh index 32b363c..f189b54 100644 --- a/download.sh +++ b/download.sh @@ -5,9 +5,9 @@ dataset_operation() { len2=$(($len1/10)) len3=$(($len2*2)) len4=$(($len3+1)) - head -n $len2 imdb_top_1000.csv.shuf > imdb_top_1000.csv.test - head -n $len3 imdb_top_1000.csv.shuf | tail -n $len2 > imdb_top_1000.csv.dev - tail -n +$len4 imdb_top_1000.csv.shuf > imdb_top_1000.csv.train + head -n $len2 imdb_top_1000.csv.shuf > imdb_top_1000_test.csv + head -n $len3 imdb_top_1000.csv.shuf | tail -n $len2 > imdb_top_1000_dev.csv + tail -n +$len4 imdb_top_1000.csv.shuf > imdb_top_1000_train.csv rm imdb_top_1000.csv.shuf wc -l imdb_top_1000.csv.* }