greedy_decorder no output about rnn-transducer HOT 4 CLOSED

It seems that your model didn't converge, the training loss is too large, you can try other optimizers like adam.

from rnn-transducer.

Have you changed the blank label index ?
You'd better check the prediction.

from rnn-transducer.

I didn't change, the blank label index is 0. Do I need to change? I checked the prediction, is all 0.

from rnn-transducer.

This is my phone map, I used character based map. The first should be the blank, index is 0.
{'': 0, 'T': 1, 'U': 2, 'S': 3, 'N': 4, 'F': 5, 'A': 6, 'C': 7, 'H': 8, 'D': 9, 'R': 10, 'W': 11, 'V': 12, 'J': 13, 'L': 14, 'Y': 15, ' ': 16, 'M': 17, 'G': 18, 'I': 19, 'O': 20, 'K': 21, 'E': 22, 'B': 23, '': 24}

from rnn-transducer.