Understanding the prediction psv file

[datduong]

Hi James,

I am trying to understand the prediction psv file. In the preds_dev.psv, why is it that each row has different length? Shouldn't each row have the same number of predicted ICD9 code (at least 15 because of the metric Precision@8 and Precision@15) ?

Thanks.