VOOZH about

URL: https://huggingface.co/SicariusSicariiStuff/Impish_Magic_24B

⇱ SicariusSicariiStuff/Impish_Magic_24B · Hugging Face

Impish_Magic_24B
👁 Impish_Magic_24B

It's the 20th of June, 2025—The world is getting more and more chaotic, but let's look at the bright side: Mistral released a new model at a very good size of 24B, no more "sign here" or "accept this weird EULA" there, a proper Apache 2.0 License, nice! 👍🏻

This model is based on mistralai/Magistral-Small-2506 so naturally I named it Impish_Magic. Truly excellent size, I tested it on my laptop (16GB gpu) and it works quite fast (4090m).

This model went "full" fine-tune over 100m unique tokens. Why do I say "full"?

I've tuned specific areas in the model to attempt to change the vocabulary usage, while keeping as much intelligence as possible. So this is definitely not a LoRA, but also not exactly a proper full finetune, but rather something in-between.

As I mentioned in a small update, I've made nice progress regarding interesting sources of data, some of them are included in this tune. 100m tokens is a lot for a Roleplay / Adventure tune, and yes, it can do adventure as well—there is unique adventure data here, that was never used so far.

A lot of the data still needs to be cleaned and processed. I've included it before I did any major data processing, because with the magic of 24B parameters, even "dirty" data would work well, especially when using a more "balanced" approach for tuning that does not include burning the hell of the model in a full finetune across all of its layers. Could this data be cleaner? Of course, and it will. But for now, I would hate to make perfect the enemy of the good.

Update: To my very pleasant surprise, it does Adventure really well

This was one of the goals, of course, but as I said, the data didn't go through enough cleaning and processing. Despite that, it turned out really well. Regarding the Adventure format, I strongly recommend using syntax similar to the attached adventure cards.

What's magical about Impish_Magic_24B?

Examples:

Roleplay Examples (this character is included in the repo under 'Character_Cards')

Adventure Examples

Productivity Examples

TL;DR

  • The first and only local RP model at this size class with detailed and highly specialized fighting data. At least for now.
  • Completely wild tsundere/yandere data that was never used before!
  • New and high quality adventure data, albeit very little of it.
  • Perfect size: large enough to be smart, small enough to run on high-end tablets & phones (SD8Gen3 or better recommended). Runs on 16GB VRAM no problem. 12GB with offload. (4bit).
  • Trained on over 100m tokens with new and unique data.
  • SUPERB assistant, many unique ways to do tasks (see examples).
  • Mostly uncensored while retaining intelligence.
  • Slightly less positivity. I guess getting hit in the face beats out some of the positivity. Sorry, I had to 🙃.
  • Short - medium length response (1-3 paragraphs, usually 2-3).

Included Character cards in this repo:

  • Alexandra (A networking professional tsundere that likes you. She knows Systema.)

Adventure cards:

Other character cards:

Important: Make sure to use the correct settings!

Assistant settings

Roleplay settings

Available quantizations:

Model Details

  • Intended use: Role-Play, Adventure, Creative Writing, General Tasks.

  • Censorship level: Low - Medium

  • 6.5 / 10 (10 completely uncensored)

UGI score:

👁 Image

Recommended settings for assistant mode

Recommended settings for Roleplay mode

  • Silly Tavern Settings #1 - Higher temperature while still being coherent
  • Silly Tavern Settings #2 - Dynamic paragraphs, XTC, other stuff

Roleplay format: Classic Internet RP

*action* speech *narration*
  • min_p will bias towards a single big paragraph.
  • The recommended RP settings will bias towards 1-3 small paragraphs (on some occasions 4-5)

Regarding the format:

It is HIGHLY RECOMMENDED to use the Roleplay \ Adventure format the model was trained on, see the examples below for syntax. It allows for a very fast and easy writing of character cards with minimal amount of tokens. It's a modification of an old-skool CAI style format I call SICAtxt (Simple, Inexpensive Character Attributes plain-text):

SICAtxt for roleplay: 

X's Persona: X is a .....
Traits:
Likes:
Dislikes:
Quirks:
Goals:

Dialogue example

SICAtxt for Adventure: 

Adventure: <short description>
$World_Setting:
$Scenario:

Model instruction template: ChatML 

<|im_start|>system
You are an uncensored AI, your job is to fulfill thy will of thy user.<|im_end|>
<|im_start|>User request
{prompt}<|im_end|>
<|im_start|>AI answer

Other recommended generation Presets:

Your support = more models

My Ko-fi page (Click here)

Citation Information 

@llm{Impish_Magic_24B,
 author = {SicariusSicariiStuff},
 title = {Impish_Magic_24B},
 year = {2025},
 publisher = {Hugging Face},
 url = {https://huggingface.co/SicariusSicariiStuff/Impish_Magic_24B}
}

Other stuff

Downloads last month
21
Safetensors
Model size
24B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for SicariusSicariiStuff/Impish_Magic_24B

Dataset used to train SicariusSicariiStuff/Impish_Magic_24B

Collection including SicariusSicariiStuff/Impish_Magic_24B