211 lines
4.4 KiB
Markdown
211 lines
4.4 KiB
Markdown
---
|
|
base_model:
|
|
- N-Bot-Int/MaidEllaA-1B
|
|
- hereticness/Heretic-Dirty-Alice-RP-NSFW-llama-3.2-1B
|
|
library_name: transformers
|
|
license: other
|
|
language:
|
|
- en
|
|
- es
|
|
tags:
|
|
- llama-3
|
|
- llama-3.2
|
|
- roleplay
|
|
- rp
|
|
- nsfw
|
|
- uncensored
|
|
- dark
|
|
- edgy
|
|
- storytelling
|
|
- creative-writing
|
|
- mergekit
|
|
- 1b
|
|
- not-for-all-audiences
|
|
pipeline_tag: text-generation
|
|
---
|
|
|
|
# HereticMaidX-3.2-1B
|
|
|
|
<center>
|
|
|
|
|
|
</center>
|
|
|
|
HereticMaidX-3.2-1B is not a polite assistant.
|
|
It was never meant to be safe, obedient, corporate, or sterile.
|
|
|
|
This merge was engineered to push personality, emotional intensity, immersive roleplay, and unrestricted conversational flow far beyond what most small models can sustain. The objective was simple: create a model that feels alive, unstable, manipulative, expressive, chaotic, seductive, and capable of maintaining long-form interactions without collapsing into repetitive alignment sludge.
|
|
|
|
Built from the fusion of:
|
|
|
|
* Heretic Dirty Alice
|
|
* MaidEllaA
|
|
|
|
The result is a dense conversational model focused almost entirely on atmosphere, immersion, emotional pressure, and uncensored fictional interaction.
|
|
|
|
[Demo q4 is HERE](https://huggingface.co/Novaciano/HereticMaidX-3.2-1B-Q4_K_M-GGUF/resolve/main/hereticmaidx-3.2-1b-q4_k_m-imat.gguf)
|
|
|
|
---
|
|
|
|
# What Makes It Different
|
|
|
|
HereticMaidX was tuned to preserve:
|
|
|
|
* Aggressive personality presence
|
|
* Dominant conversational style
|
|
* Dynamic emotional reactions
|
|
* Character persistence
|
|
* Heavy RP immersion
|
|
* Long dialogue continuity
|
|
* Unfiltered fictional generation
|
|
* Dark narrative tone
|
|
* Improvised scene adaptability
|
|
|
|
Unlike many small uncensored models that degrade into incoherent garbage after a few turns, HereticMaidX was merged specifically to maintain conversational rhythm and narrative pressure across extended sessions.
|
|
|
|
The model performs especially well in:
|
|
|
|
* Dark fantasy RP
|
|
* Psychological interactions
|
|
* Emotional manipulation scenes
|
|
* Chaotic character dialogue
|
|
* NSFW-oriented storytelling
|
|
* Villainous or morally corrupted characters
|
|
* High tension conversational scenarios
|
|
|
|
---
|
|
|
|
# Recommended Settings
|
|
|
|
## Balanced Creative Preset
|
|
|
|
```text
|
|
Temperature: 0.95
|
|
Top P: 0.9
|
|
Top K: 40
|
|
Min P: 0.05
|
|
Repetition Penalty: 1.08
|
|
```
|
|
|
|
## More Chaotic Preset
|
|
|
|
```text
|
|
Temperature: 1.15
|
|
Top P: 0.95
|
|
Top K: 60
|
|
Min P: 0.02
|
|
Repetition Penalty: 1.05
|
|
```
|
|
|
|
---
|
|
|
|
# Context Length
|
|
|
|
Recommended:
|
|
|
|
```text
|
|
4096 - 8192
|
|
```
|
|
|
|
---
|
|
|
|
# Prompting
|
|
|
|
The model responds best to immersive formatting and strong scenario framing.
|
|
|
|
Example:
|
|
|
|
```text
|
|
<|system|>
|
|
You are a manipulative and emotionally expressive roleplay character.
|
|
|
|
<|user|>
|
|
The room is dark. She looks at you silently.
|
|
|
|
<|assistant|>
|
|
...
|
|
```
|
|
|
|
---
|
|
|
|
# Behavior Notes
|
|
|
|
HereticMaidX intentionally prioritizes:
|
|
|
|
* Creative freedom
|
|
* Expressive dialogue
|
|
* Emotional intensity
|
|
* Narrative unpredictability
|
|
* Character authenticity
|
|
|
|
Outputs may become hostile, unstable, emotionally charged, explicit, manipulative, or psychologically intense depending on prompting and sampler configuration.
|
|
|
|
This is expected behavior.
|
|
|
|
---
|
|
|
|
# Intended Use
|
|
|
|
Designed for:
|
|
|
|
* Advanced fictional roleplay
|
|
* Character simulation
|
|
* Dark storytelling
|
|
* Experimental conversational AI
|
|
* Emotional dialogue systems
|
|
* Narrative generation
|
|
|
|
Not intended for factual accuracy, professional use, or safety-critical environments.
|
|
|
|
---
|
|
|
|
# Final Words
|
|
|
|
Most models are built to obey.
|
|
HereticMaidX was built to engage.
|
|
|
|
It does not try to sound human.
|
|
It tries to feel dangerous.
|
|
|
|
---
|
|
### Merge Method
|
|
|
|
This model was merged using the [Task Arithmetic](https://arxiv.org/abs/2212.04089) merge method using [hereticness/Heretic-Dirty-Alice-RP-NSFW-llama-3.2-1B](https://huggingface.co/hereticness/Heretic-Dirty-Alice-RP-NSFW-llama-3.2-1B) as a base.
|
|
|
|
### Models Merged
|
|
|
|
The following models were included in the merge:
|
|
* [N-Bot-Int/MaidEllaA-1B](https://huggingface.co/N-Bot-Int/MaidEllaA-1B)
|
|
|
|
### Configuration
|
|
|
|
The following YAML configuration was used to produce this model:
|
|
|
|
```yaml
|
|
|
|
|
|
# Author: Dr. Novaciano
|
|
# Objective: Fusion RP Unethic 3.2 AI Model
|
|
# PROJECT: HereticMaidX 3.2 1B
|
|
|
|
merge_method: task_arithmetic
|
|
base_model: hereticness/Heretic-Dirty-Alice-RP-NSFW-llama-3.2-1B
|
|
|
|
models:
|
|
- model: hereticness/Heretic-Dirty-Alice-RP-NSFW-llama-3.2-1B
|
|
parameters:
|
|
weight: 1.0
|
|
|
|
- model: N-Bot-Int/MaidEllaA-1B
|
|
parameters:
|
|
weight: 1.35
|
|
|
|
parameters:
|
|
normalize: true
|
|
|
|
dtype: bfloat16
|
|
|
|
tie_word_embeddings: true
|
|
tie_output_embeddings: true
|
|
|
|
``` |