🚀 Feature request

Add better error message to HubertForCTC, Wav2Vec2ForCTC if labels are bigger than vocab size.

Motivation

Following this issue: huggingface/transformers#12264 it is clear that an error message should be thrown if any of the any of the labels are > self.config.vocab_size or else silent errors can sneak into the training script.

So w

The Split class accepts SplitDelimiterBehavior which is really useful. The Punctuation however always uses SplitDelimiterBehavior::Isolated (and Whitespace on the other hand behaves like SplitDelimiterBehavior::Removed).

impl PreTokenizer for Punctuation {
    fn pre_tokenize(&self, pretokenized: &mut PreTokenizedString) -> Result<()> {
        pretokenized.split(|_, s| s.spl

From paper, it mentioned

Instead, the training data generator chooses 15% of tokens at random, e.g., in the sentence my
dog is hairy it chooses hairy.

It means that 15% of token will be choose for sure.

From https://github.com/codertimo/BERT-pytorch/blob/master/bert_pytorch/dataset/dataset.py#L68,
for every single token, it has 15% of chance that go though the followup procedure.

A web crawler was added by #775, but the test cases are missing.

Issue to track tutorial requests:

• Deep Learning with PyTorch: A 60 Minute Blitz - #69
• Sentence Classification - #79