Feedback
I'm currently using it with 16gb vram on sillytavern and it works like a charm. Your model is actually the first one i tried after moving from online service and although this bot is 3 times smaller, I like some aspects of it better. The 'plot armor' of user is much weaker and the bot actually managed to kill the character I rp'd as, which is very refreshing and makes stories more interesting. It follows character cards and lorebooks well and with some guidance it does alright with multiple characters and I don't mean group chats, but creating and roleplaying side characters like salesmen, guards and so on. The biggest drawback is, that it rushes the plot quite a lot. For example journey from one place to another is skipped or shortly summarized, which ruins the immersion for me. I tried to fix it by making the 'max new token' smaller, but it didn't help much. On the other hand I'm new to sillytavern so it may be fault of my settings, so I would be very thankful, if you shared your prompt.
Thank you so much for the feedback :). As for specific settings for model to develop story slowly. I havent found one that works always. Model seems to like more scifi than D&D or horror stories, I use authors note: develop the story slowly, don't rush events, introduce new obstacles in orderly manner - but it doesn't always work unf.