search

erew123 / alltalk_tts

AllTalk is based on the Coqui TTS engine, similar to the Coqui_tts extension for Text generation webUI, however supports a variety of advanced features, such as a settings page, low VRAM support, DeepSpeed, narrator, model finetuning, custom models, wav file maintenance. It can also be used with 3rd Party software via JSON calls.
GNU Affero General Public License v3.0
816 stars 91 forks source link

Documentation refers to ./start_finetuning.sh but this file does not exist #177

Closed Xyem closed 4 months ago

Xyem commented 4 months ago

Looking to try out finetuning the model on Linux and the documentation refers to ./start_finetuning.sh but this file does not exist.

erew123 commented 4 months ago

Hi @Xyem

Are you running this as a standalone installation or part of text-gen-webui? And if its a standalone, did you run the atsetup.sh to create the python environment, as that's what creates the start_finetuning.sh file.

Thanks

Xyem commented 4 months ago

Standalone and I manually created the environment (I think there is a specific reason I didn't use atsetup.sh but I can't recall what that was at the moment). Thanks for explaining why it is missing.

I did manually start the finetune (via python finetune.py) but then found out it starts a webui that you can't access remotely. I modified the script so I could.. and then found out finetuning requires 14GB of VRAM and I only have 12GB, so I can't finetune anyway. Might be worth adding a note about the 14GB requirement to the documentation? It's not mentioned in the "you will need" list.

erew123 commented 4 months ago

Hi @Xyem

Yes unfortunately Linux doesn't expand the memory over into system ram, unlike windows.

Re-documentation, I understand what you are saying. There are just so many places to try notify/document these things. I am working on a different version of AllTalk, so will look to have all the additional things in the documentation then. Ill add a quick note into the current readme though.

Thanks