SevenTailCat

Hi, do you have a solution now? After google this question, I find this is a problem about version of those python packets. however, I do not know the right...

> > my suggestion: i think you should freeze encoder, finetune ctc and decoder. > > yes, it works. but the result is worse than just using pretrained model directly....

> > you just need a bigger dataset... to expand word units.txt, you can try freeze all weight, but update the weight of your new units only. > > Hi,...

Hi, have you meet the infinitive problem?(it means that the loss or the covariance_matrices will be infinite and will assert `torch.isfinite(covariance_matrices).all()` If it is possible, could you share the solution?...

You should open the micphone in the chrome.

hi, I meet the same question now. Do you find the solution? Could you help me, please please^ _ ^ I have tried some solutions, like setting a proxy, get...