diff --git a/Jenkinsfile b/Jenkinsfile index 9e7b9e8..96e3b2f 100644 --- a/Jenkinsfile +++ b/Jenkinsfile @@ -41,7 +41,7 @@ pipeline { sh 'kaggle datasets list' } } - sh './download.sh' + sh './download.sh' $CUTOFF archiveArtifacts artifacts: 'all_games.csv.dev, all_games.csv.test, all_games.csv.train', followSymlinks: false sh 'head -n $CUTOFF all_games.csv.train' } diff --git a/download.sh b/download.sh index b04aa89..9a283d1 100755 --- a/download.sh +++ b/download.sh @@ -1,5 +1,6 @@ dataset_operation() { - cat all_games.csv | shuf > all_games.csv.shuf + cat all_games.csv | shuf > all_games.csv.s + head -n $CUTOFF all_games.csv.s > ./all_games.csv.shuf len1=$(cat ./all_games.csv.shuf | wc -l) len2=$(($len1/10)) len3=$(($len2*2))