# machine-learning
- p
pi
05/25/2021, 5:40 PMI suggested an arpabet substitution scheme in the mean time - p
pi
05/25/2021, 5:40 PMso maybe they did it like that - p
pi
05/25/2021, 5:40 PM🤷 - r
RGM - ERIOL V2 MODEL ON DEV
05/25/2021, 6:06 PMhi guys, so yeah, i'm getting help from lu9 - m
mega b
05/25/2021, 6:06 PM👍 - j
JacobLenstar
05/25/2021, 6:07 PM"why don't you take a seat" - j
JacobLenstar
05/25/2021, 6:07 PMzorak sounds like he has one of those artificial voice boxes from decades of smoking - a
Awesome Face (GTX 1660 Ti)
05/25/2021, 6:07 PM😂 omg - a
Awesome Face (GTX 1660 Ti)
05/25/2021, 6:08 PMvocoder bein SMOKER from l4d - a
Ask for dataset = EATEOT
05/25/2021, 6:09 PMI wonder if Viridi would sound better since she has more files - u
user
05/25/2021, 7:56 PMGonna try my first 30 wavs model (Freddie Mercury) - u
user
05/25/2021, 8:17 PMI've tried to downscale the batch size but it's not working - p
pi
05/25/2021, 8:17 PMyou need to force free the cuda memory - p
pi
05/25/2021, 8:18 PMeasiest for you is probably just to restart the runtime though - u
user
05/25/2021, 8:18 PMokay gonna restart it again - u
user
05/25/2021, 8:19 PMSame error - p
pi
05/25/2021, 8:19 PMwhat did you do exactly? - p
pi
05/25/2021, 8:20 PMtorch.cuda.empty() - p
pi
05/25/2021, 8:20 PMis should force free cuda memory - u
user
05/25/2021, 8:20 PMOkay gonna try it - m
mega b
05/25/2021, 8:20 PMYou need to restart runtime when this happens - m
mega b
05/25/2021, 8:20 PMor do that - u
user
05/25/2021, 8:21 PMI've restart it and doing the setup again (like initialising) - p
pi
05/25/2021, 8:21 PMCopy codetorch.cuda.empty_cache() torch.cuda.empty() - u
user
05/25/2021, 8:24 PMI need to create a new code block right? - p
pi
05/25/2021, 8:24 PMyes - u
user
05/25/2021, 8:25 PMI guess my colab is broken - p
pi
05/25/2021, 8:27 PMthat's probably because of the pytrch version - p
pi
05/25/2021, 8:27 PMit's probably fine - u
user
05/25/2021, 8:30 PMStill not working