How continue the training in google colab

[xkixkio]

I have a question about continuing training in RVC.
I trained my model with google colab for let's say 100 epochs, but I had to stop. Now I want to continue training the next day from where I left off. How can I do that?
Do I just start a new training session with the same settings? Or is there a specific way to load the checkpoint and continue from epoch 100?
Any help would be appreciated! Thanks!

[JackismyShephard]

There is an option to clear previous training attempt (don't recall right now what it's called ). if that is not set you will just continue from your last attempt. But remember to set your num epochs to be higher than the final epoch from your last attempt. It will continue from your last epoch and train until the desired epoch num has been reached.

However, if you are starting a new session on colab then there will be no existing data. So you cannot continue training in that case I think. You would have to manually repopulate the data which is not easy.

At one point I tried getting colab to work with existing data that would be mounted from your gdrive but I could not get that to work with all the settings and data it needs to work with.

You are welcome to open a pr with a solution if you know of one.

[juncgolcar-svg]

I've had an issue with error messages at the end of the training, after 8 hours or so

[JackismyShephard]

@juncgolcar-svg What error message? please raise a new issue with specific details 😄