Chatbots	Settings	Guides	Index

Sukino's SillyTavern Settings and Presets

These are the settings I use for AI Roleplaying, and with which I test all my bots. Want to know more about why I set things up the way I do? Check out my Guides page. Want to learn more? Check out my Findings page, where I share things that even smarter people have discovered too.

Banned Tokens for KoboldCPP

Are you tired of ministrations sending shivers down your spine? Do you swallow hard every time their eyes sparkle with mischief and they murmur to you barely above a whisper? Maybe, just maybe, I have the solution for you!

This is a list of clichés and repetitive phrases to ban from your AI’s vocabulary using KoboldCPP's Anti-Slop feature. Try it out, it's easy to undo if you don't like it.

Download

Presets for Text Completion Connections

Game Master Mode — The idea behind this prompt is to frame the session as a simulation rather than a roleplay, and to have the AI act as the narrator/game master rather than playing a specific character.
Jailbreak for Gemma 2 9B IT — This is the method I found to get around Gemma 2's refusals.
Conversions — These are conversions I made from other users' presets to a different instruct template:
- Methception 1.4.3 — Mistral V7
- momoura's peepsqueak 0.52 — Deepseek Chat and Reasoner
- pixi's weep 4.1 — Deepseek Reasoner

What I'm Using

Software

KoboldCPP: https://github.com/LostRuins/koboldcpp — My backend of choice, use it to load the AI models.
SillyTavern: https://github.com/SillyTavern/SillyTavern — My frontend of choice, the bridge between you and the AI model. Load your characters in it, connect to KoboldCPP and start chatting.
MikuPad: https://lmg-anon.github.io/mikupad/mikupad.html — Need to talk directly to the model without characters and system prompts in the way? Use this. I use it all the time to test how the model talks and responds to prompts on its own, and what it can or can't do.
ComfyUI: https://github.com/comfyanonymous/ComfyUI / https://github.com/ltdrdata/ComfyUI-Manager — My image generation software of choice. Node based, not really beginner friendly, but it's pretty powerful, you have control over every single generation step.

Text Generation Models

Mistral Small 3.1 24B Instruct 2503: https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Instruct-2503 — Are these Small models really 20B? They feel a lot smarter than they should. But the prose is pretty bland, and it needs a well thought-out prompt to RP well. I prefer blandness to sacrificing smartness for a finetune.
Gemma 2 9B IT: https://huggingface.co/google/gemma-2-9b-it — It's an impressive model, has a fantastic prose for such a small size, feels better than even 12B models. But it's pretty censored by default, you'll need to jailbreak it for not-so-wholesome RPs. And don't go beyond 12K context, the model will break.
Cydonia-v1.2-Magnum-v4-22B: https://huggingface.co/knifeayumu/Cydonia-v1.2-Magnum-v4-22B — Magnum V2 and V3 were great models, while V4 was kind of a miss. But for some reason, merging it with Cydonia made Cydonia even better? Go figure.
MN 12B Mag Mell R1: https://huggingface.co/inflatebot/MN-12B-Mag-Mell-R1 — Best 12B model for RP to date, imo, as smart as a 12B Mistral based model can be.
Wayfarer 12B: https://huggingface.co/LatitudeGames/Wayfarer-12B — Curious model, not the best for general RP, but shines if you like the AI Dungeon format, it was trained for that.

Image Generation Models

ΣΙΗ: https://civitai.com/models/1217645/sih — What the hell, how do you read that? I just call it sigma. Another NoobAI merge. Still testing it, but I feel like it's a big improvement over CyberFix, it's taken its place for now.
NoobAICyberFix: https://civitai.com/models/913998/noobaicyberfix — NoobAI is one of the latest efforts for anime-style, could easily replace Pony as the standard. This CyberFix makes it easier to work with.