# machine-learning
- m
mega b
05/26/2021, 4:35 AMhow much training data - m
mega b
05/26/2021, 4:35 AMthis may sound better on the site, though - s
SuperLuig
05/26/2021, 4:36 AMyeah but i'm talking about the "having a complete seizure trying to pronounce basic words" part - m
mega b
05/26/2021, 4:36 AMlol - m
mega b
05/26/2021, 4:36 AMsend the dataset here - m
mega b
05/26/2021, 4:36 AMthis should be much straighter - m
mega b
05/26/2021, 4:37 AMfor that long of training - s
SuperLuig
05/26/2021, 4:37 AMyeah its odd - s
SuperLuig
05/26/2021, 4:38 AMoh okay, so NOW it decides to produce a working result??? - s
SuperLuig
05/26/2021, 4:40 AMhttps://drive.google.com/drive/folders/19KUFV11jY0VvctBCu4KGAyDvw5imdHvZ?usp=sharing here's the dataset anyways - b
bericks
05/26/2021, 4:42 AMit's recommend that the validation loss should be somewhere below 0.1 if you want good quality - s
SuperLuig
05/26/2021, 4:47 AMlike how does the character with literally two minutes of data sound better than this lol - b
bericks
05/26/2021, 4:49 AMi trained alot for the jimmy hopkins voice, i waited till the vaildation loss was 0.1 something - s
SuperLuig
05/26/2021, 4:49 AMvalidation loss is below 0.1 - b
bericks
05/26/2021, 4:49 AMoh - s
SuperLuig
05/26/2021, 4:50 AMi dont think i can salvage the retraining unfortunately - b
bericks
05/26/2021, 4:51 AMdamn - s
SuperLuig
05/26/2021, 4:51 AMunless someone else is willing to look into it - s
SuperLuig
05/26/2021, 5:13 AMhmm - s
SuperLuig
05/26/2021, 5:31 AMokay i think i can salvage it after all - s
SuperLuig
05/26/2021, 5:31 AMsomehow the voice quality skyrocketed - s
SuperLuig
05/26/2021, 5:34 AMretrain - s
SuperLuig
05/26/2021, 5:35 AMcurrent model - s
SuperLuig
05/26/2021, 5:35 AMmfw it actually worked - s
SuperLuig
05/26/2021, 5:39 AMok definitely submitting this model - u
user
05/26/2021, 6:32 AMAre the free dots supported? - u
user
05/26/2021, 6:33 AMBecause I've tried to train my model, the validation loss wasn't decreasing at all and the curve was really deteriorated - p
pi
05/26/2021, 8:46 AMwhat are "free dots"? - u
user
05/26/2021, 8:48 AMthree dots - u
user
05/26/2021, 8:49 AMI've tried to remove them but it seems to be broken again (validation loss is not decreasing):