# machine-learning
- p
pi
06/14/2021, 5:18 AMalignment is the better proxy for output quality here - v
VeryThickBoy
06/14/2021, 5:18 AMits still robotic' - p
pi
06/14/2021, 5:19 AMwhat size is the training dataset - v
VeryThickBoy
06/14/2021, 5:19 AM143 audio clips - p
pi
06/14/2021, 5:19 AMis all audio in it high quality? - v
VeryThickBoy
06/14/2021, 5:19 AMyes it is - p
pi
06/14/2021, 5:19 AMyou're sure? - v
VeryThickBoy
06/14/2021, 5:19 AMit took me a while to find good interviews - v
VeryThickBoy
06/14/2021, 5:19 AMand im sure - v
VeryThickBoy
06/14/2021, 5:20 AMits the best i could find - p
pi
06/14/2021, 5:20 AMhow many epochs? - v
VeryThickBoy
06/14/2021, 5:20 AMshit im don't remember - v
VeryThickBoy
06/14/2021, 5:20 AMis there a way to check it after you end the training run - p
pi
06/14/2021, 5:21 AMno - v
VeryThickBoy
06/14/2021, 5:21 AMwell, it was getting up there - v
VeryThickBoy
06/14/2021, 5:21 AMi set it to run for 500 all together - p
pi
06/14/2021, 5:21 AMexcept reading the tfout file - p
pi
06/14/2021, 5:21 AMbut I assume thats beyond your capabilities - v
VeryThickBoy
06/14/2021, 5:21 AMwhere's the file located? - p
pi
06/14/2021, 5:21 AMdrive - v
VeryThickBoy
06/14/2021, 5:21 AMokay give me a second to find it - p
pi
06/14/2021, 5:22 AMwhy bother - v
VeryThickBoy
06/14/2021, 5:22 AMdunno, just give me advice thats all i need - p
pi
06/14/2021, 5:22 AMyou cant read it anyway I assume - v
VeryThickBoy
06/14/2021, 5:22 AMyeah - p
pi
06/14/2021, 5:22 AM🤷 - v
VeryThickBoy
06/14/2021, 5:22 AMhm - v
VeryThickBoy
06/14/2021, 5:22 AMill just keep running it, the graph gets better about every 20ish minutesd - p
pi
06/14/2021, 5:22 AMin most cases the audio files used for training arent good enough - p
pi
06/14/2021, 5:23 AMso if you link them I'd take a listen