# machine-learning
- w
watertail
11/05/2021, 6:28 PM - w
watertail
11/05/2021, 6:28 PMthis? - u
(Dawn) Will Draw Fictional Women
11/05/2021, 6:30 PMyeah - w
watertail
11/05/2021, 6:43 PMCUDA oom at 600 epochs with a p100 - w
watertail
11/05/2021, 6:43 PM - w
watertail
11/05/2021, 6:44 PMcolab pro btw 10 dollars a month btw - w
watertail
11/05/2021, 6:44 PMjust 79 files also - u
(Dawn) Will Draw Fictional Women
11/05/2021, 6:47 PMwhats your batch size - w
watertail
11/05/2021, 6:48 PM79 .wav files, in total 21mb - u
(Dawn) Will Draw Fictional Women
11/05/2021, 6:48 PMno i mean - u
(Dawn) Will Draw Fictional Women
11/05/2021, 6:49 PMwhat did you set your batch size to - w
watertail
11/05/2021, 6:49 PM30 - u
(Dawn) Will Draw Fictional Women
11/05/2021, 6:49 PMhuh - u
(Dawn) Will Draw Fictional Women
11/05/2021, 6:49 PMidk then - u
(Dawn) Will Draw Fictional Women
11/05/2021, 6:49 PMi can train just fine with upwards from 30 - w
watertail
11/05/2021, 6:49 PMi believe this is the exact same issue i had before when i was allocated a k80 and decided to just try 30 and it worked - w
watertail
11/05/2021, 6:51 PMcould i be misunderstanding the issue? - m
mega b
11/05/2021, 6:57 PMuse #841437191073955920 - m
mega b
11/05/2021, 6:57 PMtop most training colab - w
watertail
11/05/2021, 6:58 PMdo you mean that i should use the uberduck tacotron 2 notebook ? - m
mega b
11/05/2021, 7:56 PMYea - u
user
11/05/2021, 8:28 PMI have trained my pyrocynical model to 250 epochs time to see how it sounds - u
user
11/05/2021, 8:30 PMoh damn it sounds great - u
user
11/05/2021, 8:31 PM"When the imposter is sus" - u
user
11/05/2021, 8:33 PMhonestly the source video is quite stable so I'm super happy with this result - u
(Dawn) Will Draw Fictional Women
11/05/2021, 8:33 PMi gotta make another youtuber voice - u
user
11/05/2021, 8:34 PM📈 - f
F.B
11/05/2021, 8:37 PMsad