Is there an instruct version, or a guide to use this model correctly?
Hello! I was wondering if there was an instruct version or a guide to run this model correctly, please, because I tried using it and just kept getting terrible results, wether in instruct mode or in story (single-user) mode. I'm kinda sad about it because the description and its training data really got my attention, I'd love to have it running correctly, as well as Mag-Mell did. Thank you! (I am using a GGUF version by the way)
Hi! Thank you for the interest <3
We're working on it I promise D: we're currently slightly struggling on GPU time and haven't had the time to test making more stable and easier to use mixes atm
The main way you would use this model is as a text completion model exclusively for stories -- in other words, paste a chunk of a chapter in and press generate a bunch of times until it generates the rest of the chapter. There is no instruct template or examples for this model because it just wouldn't work.
We have a Discord if you want to join there to talk about it faster than an HF discussion :3
I'll definitely keep an eye at the progress, the model looks very promising!