• 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
ecker commented on issue ecker/ai-voice-cloning#41 2023-02-28 01:23:41 +00:00
Assorted feedback and some questions

Doesn't seem I can give specific access to the wiki, I believe I'd have to grant you access as a collaborator for this repo entirely.

I'll get around to rewriting most of it eventually.

ecker commented on issue ecker/ai-voice-cloning#39 2023-02-28 01:11:03 +00:00
Is it possible to enable saving model / training_state mid-epoch?

It would be great if there was a way to remove older checkpoints and training states during training, as it can eat up disk space quickly.

Sorry for the wait, I just happened to incidentally…

ecker closed issue ecker/ai-voice-cloning#39 2023-02-28 01:11:03 +00:00
Is it possible to enable saving model / training_state mid-epoch?
ecker pushed to master at ecker/ai-voice-cloning 2023-02-28 01:08:17 +00:00
fda47156ec oops
ecker pushed to master at ecker/ai-voice-cloning 2023-02-28 01:04:47 +00:00
bc0d9ab3ed added graph to chart loss_gpt_total rate, added option to prune X number of previous models/states, something else
ecker commented on issue ecker/ai-voice-cloning#46 2023-02-27 19:27:58 +00:00
Getting error "cuDNN error: CUDNN_STATUS_MAPPING_ERROR"

I tried the --low-vram command but I still get the error.

The Low VRAM flag is for hinting to keep models not directly being used off the GPU during generation and I think disables some…

ecker pushed to master at ecker/ai-voice-cloning 2023-02-27 19:20:20 +00:00
6925ec731b I don't remember.
ecker pushed to master at ecker/ai-voice-cloning 2023-02-27 19:11:22 +00:00
2a4c6ca193 I don't remember.
ecker commented on issue ecker/ai-voice-cloning#46 2023-02-27 17:10:35 +00:00
Getting error "cuDNN error: CUDNN_STATUS_MAPPING_ERROR"

K80 my homelab GPU is on cuda 11.4.

Ah, that could be it. The setup script uses PyTorch with CUDA 11.6 11.7.

I believe you can run the following to uninstall the wrong version and…

ecker commented on issue ecker/ai-voice-cloning#46 2023-02-27 14:11:46 +00:00
Getting error "cuDNN error: CUDNN_STATUS_MAPPING_ERROR"

Seems more like an internal CUDA error rather than something I can directly deal with. Outside of reinstalling pytorch and what-not, I don't really have any ideas.

ecker commented on issue ecker/ai-voice-cloning#44 2023-02-27 14:09:41 +00:00
Feature Request: tortoise-fast-tts

No.

I already leverage my own fork: https://git.ecker.tech/mrq/tortoise-tts.

ecker closed issue ecker/ai-voice-cloning#44 2023-02-27 14:07:50 +00:00
Feature Request: tortoise-fast-tts
ecker commented on issue ecker/ai-voice-cloning#43 2023-02-27 03:28:34 +00:00
"ModuleNotFoundError: No module named 'tortoise'"

Make sure there's a tortoise folder under ai-voice-cloning, and re-run the update script, it'll try to install tortoise.

ecker commented on issue ecker/ai-voice-cloning#41 2023-02-26 18:56:21 +00:00
Assorted feedback and some questions

Alright, I'm relatively awake now and should have some time.

I've combed through the wiki and cleaned it up to address the feedback, I guess there were a lot of outdated info than I remember. I…

ecker commented on issue ecker/ai-voice-cloning#40 2023-02-26 17:48:15 +00:00
Could not find module torchaudio\lib\libtorchaudio.pyd

So this entire time it was because I was using the older version of CUDA PyTorch. Using cu117 works for python3.10, and also works on python3.9. I guess that solves the mystery on what exactly is…

ecker closed issue ecker/ai-voice-cloning#40 2023-02-26 17:48:15 +00:00
Could not find module torchaudio\lib\libtorchaudio.pyd
ecker pushed to master at ecker/ai-voice-cloning 2023-02-26 17:47:06 +00:00
47abde224c compat with python3.10+ finally (and maybe a small perf uplift with using cu117)
ecker commented on issue ecker/ai-voice-cloning#40 2023-02-26 15:14:19 +00:00
Could not find module torchaudio\lib\libtorchaudio.pyd

Someone else mentioned having a similar problem but didn't note his python version, and he noted installing the 11.7 version fixed it:

python -m pip uninstall torch torchvision torchaudio
py…
ecker closed issue ecker/ai-voice-cloning#42 2023-02-26 15:01:40 +00:00
Add persistent_workers options in DataLoader to make training faster by removing pauses between epochs.
ecker pushed to master at ecker/DL-Art-School 2023-02-26 14:56:35 +00:00
71cc43e65c added a flag (thanks gannybal)