Loss rate:

I used V1 because V2 is still experimental.

My dataset was trained with off the wall + thriller. Did you add BAD era acapellas into the dataset?

I've used an dataset that was on post-your-dataset channel

Yeah mine was off the wall + thriller

Hehe, at least i make my best effort to make models.

Even tho i need some advices to get good acapellas

These were the tracks you used to make your MJ model?

Yes

You know how to get good acapellas with UVR?

I use multitrack acapellas for my models

i'm using UVR to make acapellas to make an dataset for an Pacho El Antifeka model i'm planning to do

Is an latin rapper that got killed yesterday

Don't Mind Me Working On A Smooth Criminal Michael Jackson

This one is good

Thanks for this great model.