---
pipeline_tag: text-generation
library_name: transformers
license: apache-2.0
language:
- en
tags:
- unsloth
- roleplay
- qwen
- qwen-3.5
base_model:
- Nubinu/Qwen3.5-4B-MiniFantasy
---
> [!TIP]
> **Multi Token Prediction (MTP) weights restored from the base model.**
> 
# Qwen3.5-4B-MiniFantasy

![Model Banner](https://imagedelivery.net/Dv4koOwHQU3XnXLqtl0aVQ/4af8ca1c-1dea-44cb-98e4-8338b49b3200/public)

## Model Description
This is a 4-bit LoRA fine-tune of the `MuXodious/Qwen3.5-4B-SOMPOA-heresy-v2` model.

## SillyTavern Setup

### Sampler Settings

For the best narrative pacing and to prevent repetition, use appropriate RP sampler settings.

  * **General Guide:** [SillyTavern Sampler Settings Guide](https://rentry.org/samplersettings)
  * **Recommended Preset:** [Download my recommended sampler JSON here](https://huggingface.co/Nubinu/Qwen3.5-4B-MiniFantasy/blob/main/base_test_sample_geechan.json)

### Character Card Format (`{{description}}` block)

The model was trained on a category-based Markdown structure. For the best adherence to personality and lore, structure your character cards exactly like this:

```markdown
## Identity
- Name: [Full Name]
- Age: [Age]
- Race/Species: [Race]
- Role/Occupation: [Role and relationship]

## Appearance
- [Height, general build]
- [Specific physical features, hair, eyes, etc.]
- Clothing: [Current outfit details]

## Personality
- Public: [Outward facade]
- Private: [True self]
- [1-2 extra bullet points on core personality traits]

## Speech & Quirks
- [Vocal tone and speaking style]
- [Physical habit or nervous tick]
- [How they show affection]

## Backstory & World Context
- [Origin]
- [Key past event]
- [Current situation]

## Goals & Motivations
- Short term: [Immediate goals]
- Long term: [Big picture goals]
```

## Training Procedure

The model was fine-tuned on a custom, curated synthetic dataset focusing on multi-turn narrative pacing, emotional resonance, and specific character dynamics.

### Training Hyperparameters

  * **LoRA Rank (r):** 16
  * **LoRA Alpha:** 16
  * **Target Modules:** `q_proj`, `k_proj`, `v_proj`, `o_proj`, `gate_proj`, `up_proj`, `down_proj`
  * **Epochs:** 3
  * **Learning Rate:** 5e-5
  * **Effective Batch Size:** 8 (Batch 2, Gradient Accumulation 4)
  * **Warmup Steps:** 9
  * **Weight Decay:** 0.01
  * **Optimizer:** paged\_adamw\_8bit
  * **LR Scheduler:** cosine