Gagan Bhatia commited on
Commit
5483f1b
1 Parent(s): 3c2d31d

Update process_data.py

Browse files
Files changed (1) hide show
  1. src/data/process_data.py +1 -0
src/data/process_data.py CHANGED
@@ -5,3 +5,4 @@ def process_data(split='train'):
5
  df = pd.read_csv('C:/Users/gbhat/Documents/GitHub/summarization/data/raw/{}.csv'.format(split))
6
  df.columns = ['Unnamed: 0', 'input_text', 'output_text']
7
  print(df.columns)
 
 
5
  df = pd.read_csv('C:/Users/gbhat/Documents/GitHub/summarization/data/raw/{}.csv'.format(split))
6
  df.columns = ['Unnamed: 0', 'input_text', 'output_text']
7
  print(df.columns)
8
+ df.to_csv('C:/Users/gbhat/Documents/GitHub/summarization/data/processed/{}.csv'.format(split))