Hey! I think it would be useful to have a more detailed explanation about:

1. what the dataset should look like for performing NER, similar to the fine-tuning example. The NER sample is great and I think it could be further improved to include an explanation about what input is expected: format of the dataset, separating sentences with newlines etc.

2. what the [start_token, end_token] mean in label_list for NER. Documentation on why we need label_list=['[PAD]', 'X', label1, label2, ...].

I could also help with the first one, based on my learnings so far :)