Model Ready ✅
This model has been synced with Weights and is ready to use for free!
1 messages · Page 1 of 1 (latest)
This model has been synced with Weights and is ready to use for free!
🫡 🦈
This shark stares at me through my window at 3 am
Thats just your grandma cosplaying
Oh no 😨
How does only 46 epochs sound good?
Klm
It works differently to normal pretrains
How so?
I dont know exactly
It just is
an hour dataset wouldnt help it
I mean it would but just a smidge
Klm is just that good
use rx 10 or 7 or 11 to clean dem dataset
Is it better than the og pretrain tho?
the original pretrain can also do good dataset reconstruction, it just that it takes longer
no, they're on pair
i've heard klm can already do almost perfect reconstruction in just 10 epochs (this doesn't mean you should be training just 10 epochs XD)
while in the og one it would take around 100
but at the end of the training, they will sound very similar to each other
so technically klm is better because it's quicker with similar result
klm is better if you have a singing dataset
og pretrain is meant for speech
that's fair
Gawr Gura [RVC v2] [RMVPE] [KLM v4.9] 46 Epochs 5014 Steps