I have a range of tokenized, training data for my model which initially looked like this :
sucks how one over can ruin everything