-
https://git.ecker.tech/ aims to provide a place to share my efforts while maintaining true ownership of my code, as I do not trust GitHub.
XMR: 4B9TQdkAkBFYrbj5ztvTx89e5LpucPeTSPzemCihdDi9EBnx7btn8RDNZTBz2zihWsjMnDkzn5As1LU6gLv3KQy8BLsZ8SG
- Joined on
2022-10-10
While mrq/tortoise-tts does have mechanisms to load a different model, there's no way to do that with just the repo itself. As I've told you before, that repo is only for TorToiSe itself and not…
You can leave it in the ./training/{voicename}-finetune/models/ folder that it came from, or you can move it to ./models/finetunes/. The model list will check for both locations.
c2726fa0d4
added new training tunable: loss_text_ce_loss weight, added option to specify source model in case you want to finetune a finetuned model (for example, train a Japanese finetune on a large dataset, then finetune for a specific voice, need to truly validate if it produces usable output), some bug fixes that came up for some reason now and not earlier
If it happens again, what can I do or check to provide a clearer picture as to what caused it?
Unfortunately, I'm not too sure how to get a good idea on what happened when it does happen,…
This may seem like a dumb question but how exactly do I get the full stack trace?
Basically everything that was printed to the console when you executed the script.
Yes, what the command…
I see on the 'Generate Configuration' tab of the web ui 'Resume State Path'. Am I correct in thinking if I paste in the most recent state path from the training_state directory it will resume…
Forgot to mention I think yesterday when I remembered to commit it: added support for it in commit 6925ec731baebf08de3fc81368559e08684c047f.
Although, I couldn't get it to compile for the life…
A full stack trace would be nice.
My guess is you somehow downloaded this repo as a ZIP and don't actually have git installed, so it's not initializing the submodules (being dlas and…
I am still getting out of memory errors with the training though, any reason why?
Besides lowering the training batch size and mega batch factor or using half-precision and crossing your…
Ah, looks like the total size of your voice inputs are too large. Increase the Voice Chunk slider and click (Re)compute Latents until it gets you something without throwing the error. I'm not…