Training on a NEW language

[rudransh2004]

Suppose we have to train this TTS model on a language and the tokens of that language are not in the Flan-T5 transformer. So can I simply change the name of the tokenizer in the config.json or do I have to make any code changes also. NOTE The new tokenizer will not be of FLAN-T5

[ylacombe]

Hey @rudransh2004, you can do this but you'd have to retrain the model from scratch!

[rudransh2004]

Hey @ylacombe thank you soo much for your reply. Could you share some reciepe for doing this with some another language without the annotations if we wish to

[ylacombe]

Hey @rudransh2004, if you want to avoid using the annotations, you could simply use a description column with each samples having empty string "". Note that the model currently doesn't support passing samples without annotations, but the trick above should work