https://uberduck.ai/ logo
Join Discord
Powered by
# machine-learning
  • p
    if you want in depth help you should share the dataset and how many epoch you trained for
  • p
    and what the final loss was
  • h
    Wonder if sentence-mixing the original then extrapolate the missing syllables would do the trick
  • p
    what?
  • h
    Probably not but that's just conjecture on my part
  • p
    you put gibberish in you get gibberish out
  • u
    I don't remember but the final loss was like 0.15, around 300 epoch and the DK rap used as the dataset
  • p
    link to the dataset?
  • u
    I don't have it but I gonna remake it in order to improve my model
  • h
    true true
  • p
    so you deleted it?
  • u
    Yeah like a dumb but it wasn't hard to do so I will do it again
  • a
    Can anyone tell me what happened?
  • a
    @User
  • a
    @User
  • h
    I think I’ve had this error before, because you changed something in the middle, you have to rerun all of the cells after the one that says “After running this cell, you must add wav files”
  • h
    It’s because the .nyp didn’t match the data you have, so you’ll have to rerun it
  • h
    Or is something else but idk lol
  • a
    ok
  • r
    I'm making Brazilian Portuguese Eriol Hiiragisawa from Cardcaptor Sakura's AI Uberduck model right now with Lu9, and he's been helping me making this model. Already had an entire audio dataset, so that's pretty cool.
  • r
    Also, my thanks to Lu9 for helping out.
  • a
    Nope not working
  • h
    Wait, I think I see the problem in the error message, you made a slight mistake in the training transcript
  • h
    I’m not sure but it does say something about the “|”
  • a
    The transcript was fine...
  • p
    send your dataset so I can take a look
  • a
  • p
    you cant have empty lines in your transcript
  • p
    delete them
  • a
    Where are they?
1...129130131...1068Latest