https://uberduck.ai/ logo
Powered by
# machine-learning
  • c
    the entire k80 thing
  • m
    you believe google cares about people?
  • h
    indirectly yes
  • h
    but only to the extent that it makes them money
  • h
    anyway
  • j
    Colab is a popular choice for the people who make quite unethical nsfw swaps, hate those people
  • h
    wonder what the performance ratio is for t4 vs, say, rtx 3090
  • u
    apparently the t4 should actually have speeds on par with an a100 i think
  • j
    Its like 1800ms vs 1100ms with p100
  • u
    its colab's alloted disk space that throttles it
  • s
    Google Colab threatens to block me from further use of Google Colab
  • j
  • j
    This is exactly what people in the dfl discord have been getting
  • c
    warning you are doing haram do you wish to continue doing haram [yes] [no]
  • h
    might straight up just start using vast.ai tomorrow
  • j
    it literally does nothing besides a warning for now though
  • r
    Where's the source code I'll fucking make this shit run on my Celeron cpu for all I care
  • m
    i haven't gotten any warnings so far and i've used tacotron training and jukebox today
  • c
    i have been training from My Computer for a while with my big scrumptious 1650
  • u
    i think it mighe be marking accounts
  • c
    it takes like the whole day
  • c
    with my tu tu tu batch size
  • j
    Obfuscate all the code and never mention the git repository or the term deepfake
  • s
    https://colab.research.google.com/github/PradyumnaKrishna/Colab-Hacks/blob/master/Colab%20RDP/Colab%20RDP.ipynb I tried to setup remote desktop connection to Google Colab backend and Colab has given me warning
  • u
    yeah they shut those down too
  • r
    Oh shit huh
  • u
    no rdp's
  • c
    port colab notebook to lipu python
  • c
    (this is what i have essentially done by setting up a local tacotron2 repository)
  • j
    Best advice is wait it out and continue training I guess
1...879880881...1068Latest