epoch1
epoch2
epoch3
epoch4
epoch5
epoch6
epoch7
epoch8
epoch9
epoch10
test result and score

best_valid_f1_score： 0.676902669973782

zip depend on the shortest

zip() takes one or more sequences and weaves their elements together, just as map(None, ...) does with sequences of equal length. The weaving stops when the shortest sequence is exhausted.

serialization序列化

i.e.接下来 e.g.例如 etc.等等

numpy的argmax求最大值的索引,从0开始
axis ˈæksɪs 轴线
axis是0-n是从索引最外层到最内层变化的
argmax(a belongs XY,axis=0)
==argmax(a[0][i],a[1][i],a[2][i],a[3][i],a[4][i]) i belongs to 0 -- Y-1
argmax(a belongs xY ,axis=1)
==argmax(a[i][0],a[i][1],a[i][2]) i belongs 0 ，X-1（比较看那个x最大，返回那个）
遍历dataframe通过itertuples遍历，item通过位置表示值，而不是通过属性
例如

for item in df.itertuples():
  print(item[0],item[1])