mangio

[vagrant hound]

how do i see the progress of the model training

[warm solstice]

Kindly explain it more

[vagrant hound]

like current progress or when is it going to end the train

[sudden gazelle]

Through the console

The black window that automatically opens when you execute go-web.bat

[vagrant hound]

thats what it shows

[warm solstice]

Yes!

[sudden gazelle]

That doesn't look good 😦

I don't think it's even training in the first place

[vagrant hound]

i pressed train model

[sudden gazelle]

Did you scroll down?

Is that all it says?

[vagrant hound]

yes

[sudden gazelle]

It should show something like that

[sudden gazelle]

@simple dome any idea what's wrong here?

[vagrant hound]

yeah ik

[warm solstice]

Does it shows anything like this

[vagrant hound]

i sent my console

[obtuse nimbus]

Ayo? @vagrant hound level 2 !!!

[sudden gazelle]

@tardy fossil

[sand sky]

@vagrant hound how long is your dataset + how powerful is your pc

Sometimes it just takes a small bit to start

[warm solstice]

Gtx 1060 3GB

[sand sky]

Same guy?

[warm solstice]

Look at it!

[tranquil sage]

damn 3 GB dataset

[warm solstice]

I think it's GPU capacity

[tranquil sage]

oh nvm u meant 3GB GPU RAM

[warm solstice]

Yeh! 👍

[tranquil sage]

did it solve yet?

[sand sky]

This is not even the same guy I think

[vagrant hound]

1 sec

[tardy fossil]

hmm? why was i pinged?

[vagrant hound]

the audio file is 20 min long

and my pc is 1060 3gb and i5 8600

[sand sky]

Yeah okay then it sounds like you should just wait a bit

If nothing changes 5 mins from now let us know

[tardy fossil]

what batch size did you put?

[vagrant hound]

its like this 40 mins already

batch size 1

[sand sky]

Okay nevermind

[vagrant hound]

i didnt touch the batch size at all

[sand sky]

Batch size 1 is super slow and also never recommended

[vagrant hound]

ah i didnt know oops

so how much should i put it on?

[sand sky]

You should restart and set it to 4

Like restart the program

Close the gui website, everything

[vagrant hound]

k

[tardy fossil]

anythging higher than 1 will not work for 3gb vram

[tranquil sage]

hey btw how much should i set fr a GPURAM size 4GB

[sand sky]

Damn

But can you even train a model well on 1

Like that literally sounds like a meme

[tardy fossil]

if the dataset is clean

[tranquil sage]

at this point just train on colab notebooks

[vagrant hound]

i tried

didnt work

[tardy fossil]

1060 is too weak for training

[vagrant hound]

this right?

[tranquil sage]

i did it yesterday and worked fine fr me

[vagrant hound]

https://colab.research.google.com/drive/1r4IRL0UA7JEoZ0ZK8PKfMyTIBHKpyhcw#scrollTo=w4wXvoez9Rce

the guides i found for it didnt include the ngrok auth token or what ever is that

so it didnt work atleast i think thats the reason

[tardy fossil]

ah yeah, it's not working because of some issue in feature extraction code, use rvc disconnected for now

[tranquil sage]

just open an account and get it for free

[vagrant hound]

this one right?

https://colab.research.google.com/drive/1XIPCP9ken63S7M6b5ui1b36Cs17sP-NS#scrollTo=Yj-_npuA_HRB

[tardy fossil]

yes

[vagrant hound]

is there a guide for that or no

[tranquil sage]

yes i used this one

[tardy fossil]

it works the same way as easygui basically

[tranquil sage]

but be sure to save ur data before ur session ends..so try less epochs

[warm solstice]

Just follow the steps

[tranquil sage]

it doesn't have that autosave functionality

[vagrant hound]

never used easygui

[tardy fossil]

it's this one

[shell osprey]

For rvcdisconnected?

Yes

https://docs.google.com/document/d/1XuxQYiqEhYrdYeCZRRLrmV_ciMKo0bV-jTCGHu_-5Cc/edit?usp=drivesdk