Tokenizing Issue Fine Tuning BERT

So I am following the documentation for Fine-Tuning here

My dataset is different, in that all of my labels are in one column. I was able to get them into a labels list in the same way, and make the two dictionaries, id2label and label2id.

I’m running into issues with the preprocess function where its returning 'KeyError: ‘trusting’" where trusting is one of my labels

I’m assuming it’s something to do with the labels_batch line, but I’m not certain.

Thanks in advance!