search

KoboldAI / KoboldAI-Client

For GGUF support, see KoboldCPP: https://github.com/LostRuins/koboldcpp
https://koboldai.com
GNU Affero General Public License v3.0
3.46k stars 747 forks source link

Question regarding the dataset source #383

Closed Megasister closed 1 year ago

Megasister commented 1 year ago

Hi there creators of Kobold. I am a freshman in this LLM thing and I am trying to find some useful datasets to finetune my personal model for CYOA-related content generation and happened to stumble upon your model Nerys. It is simply fascinating!

So about the Pike, CYS, and manga datasets mentioned, are those publicly available datasets or proprietary? Is there any chance I can find them anywhere?

Kind regards Larry

Megasister commented 1 year ago

My email is larryyanthrottle@gmail.com in case you want to get in touch any other way~

henk717 commented 1 year ago

Depwnds on the community tuner if they release their data publically or not. Nerys is mostly closed source (All Seekers models are since he uses them to get work in the industry by selling tailor made versions to fund new free public models) but the CYOA part you are looking for is from Skein which does have an open datsset.

You can find skeins data here : https://wandb.ai/ve-forbryderne/skein/runs/files/files/datasets

Megasister commented 1 year ago

Thanks so much. That helps a lot