Tabular model uses deep_var as an input

Hi, I have a trained NN which uses deep_var as an input. As when I’m trying to deploy it, In the input dataset and the deep_var column = 0 and the prediction is wrong, however, if I replace it with a correct number it predicts very close to that number.

Creating data in the NN trainer:
data = (TabularList.from_df(train_set, cont_names=cont_names, procs=procs)
.split_none()
.label_from_df(cols=dep_var, label_cls=FloatList, log=True)
.databunch(bs = 500))

Predicting the deep_var:
pred1 = learn1.predict(row)
print(“Predicted: {}”.format( 2.718281**(float(pred1[1]))))

Please help

Any particular reason why you did this? The only instance this could apply is if you’re using the previous value, as otherwise your model is essentially cheating the entire time. Otherwise you would treat it seperately.