🚀 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

Is there a way to train a bidirectional RNN (like LSTM or GRU) on trax nowadays?

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.

1.希望可以把底层的api文档再完善一些，比如encoder,decoder，以便于复现一些论文
2.希望可以维护一个pytorch和paddle的api对照一览表，尽量全一些
3.错误日志能否准确一些，有时候datalaoder出的一些错误信息不好定位
4.能否增加使用梯度累加特性，进一步提高batch size

Till now, we have released the English model which is trained on
academic datasets. We are planning to support recognition models of more languages.

If you want to support a new language, please provide us with two files:

1. A char_list.txt file, which lists all the characters that used in the new language.
2. A dict_list.txt file, which lists the words in the new language as more as po

Hi, I am so interesting in your project, and wonder if you need contributor and how could I make my own contribution?