nice , finally a model with 2 min dataset but not freaking 200 epochs or something

was it over kill to train it up to 1400 epochs? maybe xD

i think so ? but its better than most models here with short datasets

respect for ur dedication

I tend to get carried away with my work😅 thanks🥰

what's wrong with a 2 minute dataset having 200 epochs?

The training curve is not existent

means? I'm not familiar w the explanation

wym?

When u train with a small dataset and lower epochs, the model doesn't train as it should, it doesn't smooth the learning curve

This what i mean

better phrasing is probably that you're not training the model as optimal as it can be

like sure the model sounds good already but it could be better

pretty sure 99% of people don't give a single FUCK if it's "as optimal as it can be"

looking at most models here i'd say yeah thats the case

Personally I just make these models for fun, trying to make it "as optimal as it can be" takes a lot of the fun out of it

like you can literally see here how more epochs could *potentially* improve quality

but its not like anyone's forcing you to do it

Personally the quality to me doesn't sound any better than models i've used with way lower epoch counts

thats why i said potentially

It's at that point where you REALLY gotta pay attention to tell anything different

ive had songs where it took a higher number of epochs to infer and make it sound good, less cracks, smoother voice

ultimately comes down to your preference