Can I see a sample dataset?

#3
by kbuwel - opened

I am very interested in language models.
I tried your model in koboldcpp.
Supporting multiple languages was very interesting.
What I liked most was that it supported my native language, but it was too awkward.
I felt like I wanted to correct the awkward part.
Can you tell me which text file you used and how?
koboldcpp has the ability to write scenarios and chat, but I'm not sure how to train.

Owner

Hello!
Unfortunately, I didn't trained on any dataset as this model is a merge.
But if you create a dataset in your own language you can probably train that on top of it with axolotl (https://github.com/OpenAccess-AI-Collective/axolotl)

Sign up or log in to comment