You can also do NLP (classification)

import paddle
import numpy as np
with open("train.tsv","r",encoding="utf-8") as f:
    data=f.readlines() 
train_data=[]