# machine-learning
- m
mega b
05/18/2021, 5:05 PMThat doesn't look good for epoch 238... - b
brilliam.de
05/18/2021, 5:27 PMis this bad for epoch 16 @User - m
mega b
05/18/2021, 5:28 PMuhh looks okay - m
mega b
05/18/2021, 5:28 PMkeep training - b
brilliam.de
05/18/2021, 5:28 PMalright - b
brilliam.de
05/18/2021, 5:28 PMi have 300 epocs that im doing - b
brilliam.de
05/18/2021, 5:28 PMa max of 300 - m
mega b
05/18/2021, 5:28 PM👍 - p
pi
05/18/2021, 5:28 PMi'd do less - b
brilliam.de
05/18/2021, 5:28 PMepoch 23 - b
brilliam.de
05/18/2021, 5:29 PMthe more epocs the better it will sound bc thats how many times the AI runs thru it right? - m
mega b
05/18/2021, 5:30 PM*how many times it runs through the batch - b
brilliam.de
05/18/2021, 5:30 PMfinally a good one lol - b
brilliam.de
05/18/2021, 5:31 PMuh oh - h
HoodyP
05/18/2021, 5:36 PMI know, but, that was before lol... Now its better, it has a decent line with loss of 0.09 - m
mega b
05/18/2021, 5:36 PMah, nice! - h
HoodyP
05/18/2021, 5:37 PMActually, I just rechecked and now it’s in the 0.08 range! - m
mega b
05/18/2021, 5:37 PMoh sweet - h
HoodyP
05/18/2021, 5:37 PMGoing strong! - m
mega b
05/18/2021, 5:38 PMsheeesh 💯 - t
Toasty
05/18/2021, 5:56 PM👀 - a
Ask for dataset = EATEOT
05/18/2021, 5:57 PMwhat voice is this? - t
Toasty
05/18/2021, 6:19 PMGoro Akechi from Persona 5. - m
mega b
05/18/2021, 6:20 PMWill style tokens be possible with sv2tts? - p
pi
05/18/2021, 6:23 PMin theory: yes - h
HoodyP
05/18/2021, 6:23 PMWelp, Google Colab decided to be a jerk and disconnected my training process, and also disable my main acc’a usage of GPU backends... - m
mega b
05/18/2021, 6:24 PMKaggle is a good alternative - h
HoodyP
05/18/2021, 6:24 PMKaggle? - m
mega b
05/18/2021, 6:24 PMyes - m
mega b
05/18/2021, 6:24 PMkaggle notebooks