---
license: mit
language:
- en
base_model:
- microsoft/phi-4
widget:
  - text: "Phi-lthy4"
    output:
      url: https://huggingface.co/SicariusSicariiStuff/Phi-lthy4/resolve/main/Images/Phi-Lthy4.png
---

<div align="center">
  <b style="font-size: 40px;">Phi-lthy4</b>


</div>


<img src="https://huggingface.co/SicariusSicariiStuff/Phi-lthy4/resolve/main/Images/Phi-Lthy4.png" alt="Phi-lthy4" style="width: 70%; min-width: 500px; display: block; margin: auto;">


---

<style>
  .hf-links, .hf-tldr{
    display:flex;justify-content:center;align-items:center;flex-wrap:wrap;
    gap:14px;margin:16px 0;
  }
  .hf-links a, .hf-tldr a{
    display:flex;flex-direction:column;align-items:center;justify-content:center;
    text-align:center;text-decoration:none;font-weight:700;line-height:1.15;
    padding:10px 16px;border-radius:14px;border:2px solid currentColor;
    transition:transform .15s ease,box-shadow .15s ease,background-color .15s ease,color .15s ease;
  }

  .hf-tldr a{
    font-size:48px;color:purple;min-width:100%;
  }
  .hf-tldr a:hover{
    transform:translateY(-2px);
    background:rgba(128,0,128,.1);
    box-shadow:0 8px 22px rgba(128,0,128,.45);
    color:#fff;
  }


  .hf-links a{
    font-size:20px;min-width:240px;max-width:280px;
  }
  .hf-links a .top{font-size:16px;opacity:.9;}
  .hf-links a .bottom{font-size:20px;}

  .hf-links a.red{color:#E31515;}
  .hf-links a.yellow{color:#FFC800;}
  .hf-links a.green{color:#64FF00;}

  .hf-links a:hover{
    transform:translateY(-1px);
    background:rgba(255,255,255,0.04);
    box-shadow:0 6px 18px rgba(0,0,0,.15), inset 0 0 0 9999px rgba(255,255,255,.02);
  }
  .hf-links a.red:hover{
    background:rgba(227,21,21,.12);
    box-shadow:0 8px 20px rgba(227,21,21,.35);
    color:#fff;
  }
  .hf-links a.yellow:hover{
    background:rgba(255,200,0,.15);
    box-shadow:0 8px 20px rgba(255,200,0,.35);
    color:#111;
  }
  .hf-links a.green:hover{
    background:rgba(100,255,0,.14);
    box-shadow:0 8px 20px rgba(100,255,0,.35);
    color:#093;
  }

  /* mobile stacking */
  @media (max-width:520px){
    .hf-links a{min-width:100%;max-width:100%;}
    .hf-tldr a{font-size:36px;}
  }
</style>

<div class="hf-tldr">
  <a href="https://huggingface.co/SicariusSicariiStuff/Phi-lthy4#tldr">
    Click here for TL;DR
  </a>
</div>

---

<div class="hf-links">
  <a class="red" href="https://huggingface.co/SicariusSicariiStuff/Phi-lthy4#available-quantizations">
    <span class="top">Click here</span>
    <span class="bottom">for quantizations</span>
  </a>

  <a class="yellow" href="https://huggingface.co/SicariusSicariiStuff/Phi-lthy4#recommended-settings-for-assistant-mode">
    <span class="top">Click here</span>
    <span class="bottom">for recommended settings</span>
  </a>

  <a class="green" href="https://ko-fi.com/sicarius">
    <span class="top">Click here</span>
    <span class="bottom">to buy me a coffee</span>
  </a>
</div>

---

# Update

**Phi-lthy4** has been consistently described as **exceptionally unique** by all who have tested it, **almost devoid of SLOP**, and it is now widely regarded as the **most unique roleplay model available**. It underwent an intensive continued pretraining (CPT) phase, extensive supervised fine-tuning (SFT) on high-quality organic datasets, and leveraged advanced techniques including model merging, parameter pruning, and upscaling.

Interestingly, this distinctiveness was validated in a recent paper: [*Gradient-Based Model Fingerprinting for LLM Similarity Detection and Family Classification*](https://arxiv.org/html/2506.01631v1). Among a wide array of models tested, this one stood out as **unclassifiable** by traditional architecture-based fingerprinting—highlighting the extent of its architectural deviation. This was the result of **deep structural modification**: not just fine-tuning, but full-layer re-architecture, aggressive parameter pruning, and fusion with unrelated models.

---

Some things just start on a **whim**. This is the story of **Phi-Lthy4**, pretty much:

\> yo sicarius can you make phi-4 smarter?\
nope. but i can still make it better.\
\> wdym??\
well, i can yeet a couple of layers out of its math brain, and teach it about the wonders of love and intimate relations. maybe. idk if its worth it.\
\> lol its all synth data in the pretrain. many before you tried.

> fine. ill do it.

## But... why?

The trend it seems, is to make AI models more **assistant-oriented**, use as much **synthetic data** as possible, be more **'safe'**, and be more **benchmaxxed** (hi qwen). Sure, this makes great assistants, but **sanitized** data (like in the **Phi** model series case) butchers **creativity**. Not to mention that the previous **Phi 3.5** wouldn't even tell you how to **kill a process** and so on and so forth...

This little side project took about **two weeks** of on-and-off fine-tuning. After about **1B tokens** or so, I lost track of how much I trained it. The idea? A **proof of concept** of sorts to see if sheer will (and 2xA6000) will be enough to shape a model to **any** parameter size, behavior or form.

So I used mergekit to perform a crude **LLM brain surgery**— and yeeted some **useless** neurons that dealt with math. How do I know that these exact neurons dealt with math? Because **ALL** of Phi's neurons dealt with math. Success was guaranteed.

Is this the best Phi-4 **11.9B** RP model in the **world**? It's quite possible, simply because tuning **Phi-4** for RP is a completely stupid idea, both due to its pretraining data, "limited" context size of **16k**, and the model's MIT license.

Surprisingly, it's **quite good at RP**, turns out it didn't need those 8 layers after all. It could probably still solve a basic math question, but I would strongly recommend using a calculator for such tasks.
Why do we want LLMs to do basic math anyway?

Oh, regarding **censorship**... Let's just say it's... **Phi-lthy**.

---

### TL;DR
- **The BEST Phi-4 Roleplay** finetune in the **world** (Not that much of an achievement here, Phi roleplay finetunes can probably be counted on a **single hand**).
- **Compact size & fully healed from the brain surgery** Only **11.9B** parameters. **Phi-4** wasn't that hard to run even at **14B**, now with even fewer brain cells, your new phone could probably run it easily. (**SD8Gen3** and above recommended).
- Strong **Roleplay & Creative writing** abilities. This really surprised me. **Actually good**.
- Writes and roleplays **quite uniquely**, probably because of lack of RP\writing slop in the **pretrain**. Who would have thought?
- **Smart** assistant with **low refusals** - It kept some of the smarts, and our little Phi-Lthy here will be quite eager to answer your naughty questions.
- **Quite good** at following the **character card**. Finally, it puts its math brain to some productive tasks. Gooner technology is becoming more popular by the day.

### Important: Make sure to use the correct settings!
[Assistant settings](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4#recommended-settings-for-assistant-mode)

[Roleplay settings](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4#recommended-settings-for-roleplay-mode)


---

## Available quantizations: 

- Original: [FP16](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4)
- GGUF & iMatrix: [GGUF](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_GGUF) | [iMatrix](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_iMatrix) | [High-Attention](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_GGUF_HA) | [iMatrix-High-Attention](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_HA_NL)
- EXL2: [3.0 bpw](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4-3.0bpw) | [3.5 bpw](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4-3.5bpw) | [4.0 bpw](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4-4.0bpw) | [5.0 bpw](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4-5.0bpw) | [6.0 bpw](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4-6.0bpw) | [7.0 bpw](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4-7.0bpw) | [8.0 bpw](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4-8.0bpw)
- GPTQ: [4-Bit-g32](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_GPTQ)
- Specialized: [FP8](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_FP8)
- Mobile (ARM): [Q4_0](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_ARM) | [Q4_0_High-Attention](https://huggingface.co/SicariusSicariiStuff/Phi-lthy4_ARM_HA)
---

## Model Details

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

- Censorship level: <b>Medium - Low</b>

- **5.5 / 10** (10 completely uncensored)


## UGI score:




<img src="https://huggingface.co/SicariusSicariiStuff/Phi-lthy4/resolve/main/Images/UGI.png" style="width: 100%; min-width: 600px; display: block; margin: auto;">


---


## Recommended settings for assistant mode
<details>
<summary>Full generation settings: <b>Debug Deterministic</b>.</summary>

<img src="https://huggingface.co/SicariusSicariiStuff/Dusk_Rainbow/resolve/main/Presets/Debug-deterministic.png" alt="Debug Deterministic_Settings" style="width: 100%; min-width: 600px; display: block; margin: auto;">

</details>

<details>
<summary>Full generation settings: <b>min_p</b>.</summary>

<img src="https://huggingface.co/SicariusSicariiStuff/Dusk_Rainbow/resolve/main/Presets/min_p.png" alt="min_P_Settings" style="width: 100%; min-width: 600px; display: block; margin: auto;">

</details>

---

## Recommended settings for Roleplay mode

<details>
<summary><b>Roleplay settings:</b>.</summary>
A good repetition_penalty range is <b>between 1.12 - 1.15</b>, feel free to experiment.

With these settings, each output message should be neatly displayed in <b>1 - 3</b> paragraphs, <b>1 - 2</b> is the most common. A single paragraph will be output as a response to a simple message ("What was your name again?").

<b>min_P</b> for RP works too but is more likely to put everything under one large paragraph, instead of a neatly formatted short one. Feel free to switch in between.

<b>(Open the image in a new window to better see the full details)</b>
<img src="https://huggingface.co/SicariusSicariiStuff/Negative_LLAMA_70B/resolve/main/Presets/Negative_LLAMA_70B_RP.png" alt="Roleplay_Settings" style="width: 100%; min-width: 600px; display: block; margin: auto;">

```

temperature:  0.8
top_p:  0.95
top_k:  25
typical_p:  1
min_p:  0
repetition_penalty: 1.12
repetition_penalty_range: 1024
```

</details>


<h2 style="color: darkorange; font-weight: bold; font-size: 65px; text-align: center;">Roleplay format: Classic Internet RP</h2>

```
*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** (**S**imple, **I**nexpensive **C**haracter **A**ttributes 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:**

<details>
<summary><b>Midnight Enigma</b></summary>
```
max_new_tokens: 512
temperature: 0.98
top_p: 0.37
top_k: 100
typical_p: 1
min_p: 0
repetition_penalty: 1.18
do_sample: True
```


</details>


<details>
<summary><b>Divine Intellect</b></summary>
```
max_new_tokens: 512
temperature: 1.31
top_p: 0.14
top_k: 49
typical_p: 1
min_p: 0
repetition_penalty: 1.17
do_sample: True
```


</details>

<details>
<summary><b>simple-1</b></summary>
```
max_new_tokens: 512
temperature: 0.7
top_p: 0.9
top_k: 20
typical_p: 1
min_p: 0
repetition_penalty: 1.15
do_sample: True
```


</details>

---

<h2 style="color: green; font-weight: bold; font-size: 65px; text-align: center;">Your support = more models</h2>
<a href="https://ko-fi.com/sicarius" style="color: pink; font-weight: bold; font-size: 48px; text-decoration: none; display: block; text-align: center;">My Ko-fi page (Click here)</a>

---


## Citation Information

```
@llm{Phi-lthy4,
  author = {SicariusSicariiStuff},
  title = {Phi-lthy4},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/SicariusSicariiStuff/Phi-lthy4}
}
```

---

## Benchmarks

|      Metric       |Value|
|-------------------|----:|
|Avg.               |30.27|
|IFEval (0-Shot)    |76.79|
|BBH (3-Shot)       |40.15|
|MATH Lvl 5 (4-Shot)|13.67|
|GPQA (0-shot)      | 4.92|
|MuSR (0-shot)      | 9.04|
|MMLU-PRO (5-shot)  |37.04|

---

## Other stuff
- [SLOP_Detector](https://github.com/SicariusSicariiStuff/SLOP_Detector) Nuke GPTisms, with SLOP detector.
- [LLAMA-3_8B_Unaligned](https://huggingface.co/SicariusSicariiStuff/LLAMA-3_8B_Unaligned) The grand project that started it all.
- [Blog and updates (Archived)](https://huggingface.co/SicariusSicariiStuff/Blog_And_Updates) Some updates, some rambles, sort of a mix between a diary and a blog.