Infinite the Jackal (Sonic Forces) [RVC V2, 100 Epochs, RMVPE]

[hardy spire]

Only reason I train on 100 Epochs is because it takes too long for 300 lol

Link: https://huggingface.co/Zeal-anton-ski/Sage/resolve/main/InfiniteTheJackal.zip?download=true

[zealous pawn]

Rate the model

To rate this model, either use /rate or click the buttons below.

[mild brook]

it takes too long? how long was the dataset :p

[hardy spire]

6 minutes
I'm new to training on RVC so I don't know that much about how long certain things take

[mild brook]

Oh! 6 minutes shouldn't take too long to get to 300 although your model doesn't sound bad at all :D

[hardy spire]

Thanks 🙏
It takes 25 minutes for me to train a model to 100 Epochs, that's why I say it takes long, for me to do 300 Epochs it would take over an hour

[iron urchin]

Well It depends on what he classifies on too long

That isn't long nvm

[mild brook]

Don't worry most models take about an hour or longer if you go to 300, at least for me bc I use decently long datasets

[odd bolt]

How many batch size do you use?

[hardy spire]

7

[odd bolt]

Yk, batch size affects speed

[odd bolt]

eh, good value i think

[hardy spire]

How much should I do for the shortest

[mild brook]

does it?

[odd bolt]

it always affected speed. high batch size, fast training due to increased paralelism (idk whats parelelism but probably some machine learning thing), and low batch size, slow training due to reduced parelelism

[odd bolt]

short datasets, use from 2 to 4

large datasets, use 16 if its 30 min or more, or if its 10 min use from 6 to 8

[iron urchin]

Imma just stick with 8 fr

[odd bolt]

Ok then

[mild brook]

is 2 even stable?

[odd bolt]

yes but only for short datasets

[mild brook]

like 5 seconds short?

or shorter?

[odd bolt]

yea, u can also use batch size 1 if possible

[wet kernel]

Model Ready ✅

This model has been synced with Weights and is ready to use for free!