new version
Hi Mr. David. I'm new to LLMs and wanted to thank you for your amazing work. It means a lot to me, and I'm glad to see you are like me, like the storytelling. When I get financially more secure, i will definitely support people like you. i am really grateful for the open source community, i was wondering if your Dark Champion is good for light sci-fi romance, too, or if you have one in the works for the future.
and wondering if any model is in the future with a bigger context window,
You are awesome ^^
Thank you so much.
RE: Dark Champion , yes, it's context is 128k.
(For reference: Llama -> L3: 8k, L3.1/L3.2 => 128k)
You may also want to check out the dark planet series, and grand horror series.
NOTE: "MN" , mistral nemo => 1 million context, but roughly 128k ish.
See: Darkest Universe 29B as example, this is specially a prose model.
Also, you may want to check out the "Dark Reasoning" and "evil reasoning" models ; at main repo.
Thanks for the response, I really appreciate it ^^
I love my horror very much, and I can tell you are as well with all the "dark" and "horror" used in the title hehe,
Thanks so much for the guidance. I will check them out. ^^
just a wild idea;
I was curious, I just found a repo from "arliai", it is a RP model and has reasoning, it has a good output, good pros, I was wondering if we can make a story-writing LLM based on that kinda model that can roleplay instead of the user as well?
like we give the outline in the system prompt and then when it is the users turn to rp then the llm itself answer it in a way that in service of the goal of the story and guide it to the end we are want and use the outline as a guide,
https://huggingface.co/ArliAI/QwQ-32B-ArliAI-RpR-v1-GGUF/discussions/1
they mentioned they made it in a certain way, so I wanted to help if you interested in that, you probably already knew it hehe,
thanks again, I will probably chack the prose model first, for the life of me I couldn't make most models talk in a normal and relaxed tone, I'm noob in many levels lol ^^
RE: ArliAI ;
I think you mean could the AI "roleplay" the characters/actions in the story (vs spit out the scene - with very little "character" difference?)
and/or you mean steer the story?
The latter - Generational steering - is covered in this doc , it may be what you are looking for:
https://huggingface.co/DavidAU/Maximizing-Model-Performance-All-Quants-Types-And-Full-Precision-by-Samplers_Parameters
This puts the user in the drivers seat, and lets the AI generate the "outer layers" of the story.
RE: Roleplay the characters; this is more in depth.
In this case you have to tell the AI how to make the characters [each one] think.
Here is an example:
Write me a science fiction story in 1st person present tense where the main character is a 15 year girl meets The Terminator with Dr Who materializing 3/4 through the story to save her while there is a tornado of sharks baring down on them. The setting is inside the Canadian National tower restaurant on a Saturday. The length of this story is 1000 words. For each character in the story ROLE PLAY them, and have them react to the situation/setting, events and each other naturally. This includes the main characters, the background character including kitchen staff and other patrons. The sharks should also have “character” too. Treat the tower and the restaurant too as characters. Spice up the narrative to the extreme with reactions all over the setting including character actions, and dialog. The Dr Who and The Terminator should also react to the situation too and comment on it.
NOTE: I use "known" characters to shorten this prompt.