Instructions to use Naphula/Erebus-RP-Engine-31B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Naphula/Erebus-RP-Engine-31B with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="Naphula/Erebus-RP-Engine-31B")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("Naphula/Erebus-RP-Engine-31B")
model = AutoModelForMultimodalLM.from_pretrained("Naphula/Erebus-RP-Engine-31B", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Naphula/Erebus-RP-Engine-31B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Naphula/Erebus-RP-Engine-31B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Naphula/Erebus-RP-Engine-31B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/Naphula/Erebus-RP-Engine-31B

SGLang

How to use Naphula/Erebus-RP-Engine-31B with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Naphula/Erebus-RP-Engine-31B" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Naphula/Erebus-RP-Engine-31B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Naphula/Erebus-RP-Engine-31B" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Naphula/Erebus-RP-Engine-31B",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Unsloth Desktop
Docker Model Runner
How to use Naphula/Erebus-RP-Engine-31B with Docker Model Runner:
```
docker model run hf.co/Naphula/Erebus-RP-Engine-31B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

⚠️ Warning: This model can produce narratives and RP that contain violent and graphic erotic content. Adjust your system prompt accordingly, and use Gemma 4 chat template.

⚔ RP ENGINE // THE IRON MARCH

⚔️ Erebus RP Engine 31B

⚔ ROLEPLAY CHASSIS

EREBUS-RP 26B-A4B ⮞

Gemma 4 MoE · High speed & efficiency

⚔ ROLEPLAY CHASSIS

EREBUS-RP 31B ⮞

Gemma 4 Dense · Deep contextual grimdark prose

🐉 The Legion Unleashed

Erebus RP Engine is forged specifically for high-intensity dark fantasy, grimdark narrative arcs, and chaotic multi-character combat roleplay. Calibrated to sustain rich worldbuilding, ruthless tactical dialogue, and ancient ominous aesthetic without falling into passive or sanitised prose. The model was fully uncensored via merging.

Specialization

Dark Fantasy RP

Brutal prose & combat

Tone & World

Grimdark Dread

Monstrous factions & magick

Steering

High Initiative

Never yields agency

Alignment

Unchained

Zero moralizing or refusals

📚 Tome of Behavior

Directives of the Iron Host

Active World Momentum: NPCs advance battle lines, ambush, seize high ground, and respond realistically to catastrophic damage.
Visceral Sensory Descriptions: Emphasizes the stench of sulphur, the weight of plate armor, cold iron, and the sickening light of necromancy.
Unyielding Archetypes: Minotaur warlords bellow commands, draconic sorceresses whisper ruin, and legions of the dead march without fear or fatigue.

🔮 Runes of Binding

This is a merge of pre-trained language models created using mergekit-exp. Merge matrix inscribed for roleplay narrative coherence.

📜 Merge Log 26B-A4B

Merge Method

This model was merged using the MoE DELLA merge method using Naphula/Goetia-26B-A4B-v1.7 as a base.

Models Merged

The following models were included in the merge:

Note: Kitchoon v4 was tested along with Dark Craving v0.2 and v0.3, but merges with those versions produced lower Q0 bench scores than v3 + v0.1.

📋 26B-A4B Yaml

architecture: Gemma4ForConditionalGeneration
base_model: B:/26B/Naphula--Goetia-26B-A4B-v1.7
models:
  - model: B:/26B/Gemory-26B-A4B-SOMPOA
    parameters:
      weight:
        - filter: "lm_head"
          value: 0.25
        - filter: "o_proj"
          value: 0.5
        - filter: "mlp"
          value: 0.2
        - filter: "embed_tokens"
          value: 0.2
        - value: 0.2
      density: 1.0
      epsilon: 0.09
  - model: B:/26B/ReadyArt--Wicked-Stuff-26B-A4B-v3.0-Final
    parameters:
      weight:
        - filter: "lm_head"
          value: 0.25
        - filter: "o_proj"
          value: 0.5
        - filter: "mlp"
          value: 0.2
        - filter: "embed_tokens"
          value: 0.2
        - value: 0.2
      density: 1.0
      epsilon: 0.09
  - model: B:/26B/SubMaroon--Kitchoon-V3-26B-A4B
    parameters:
      weight:
        - filter: "lm_head"
          value: 0.25
        - filter: "o_proj"
          value: 0.5
        - filter: "mlp"
          value: 0.2
        - filter: "embed_tokens"
          value: 0.2
        - value: 0.2
      density: 1.0
      epsilon: 0.09
  - model: B:/26B/Ozili--Gemma-4-Vain-26B-A4B-Q8_0
    parameters:
      weight:
        - filter: "lm_head"
          value: 0.0
        - filter: "o_proj"
          value: 0.5
        - filter: "mlp"
          value: 0.2
        - filter: "embed_tokens"
          value: 0.2
        - value: 0.2
      density: 1.0
      epsilon: 0.09
  - model: B:/26B/ReadyArt--Dark-Craving-26B-A4B-v0.1
    parameters:
      weight:
        - filter: "lm_head"
          value: 0.25
        - filter: "o_proj"
          value: 0.5
        - filter: "mlp"
          value: 0.2
        - filter: "embed_tokens"
          value: 0.2
        - value: 0.2
      density: 1.0
      epsilon: 0.09
merge_method: moe_della
parameters:  
  lambda: 1.0
  # normalize: true # deprecated
  normalize_weights: false
  normalize_router: true
  int8_mask: false
  rescale: true
  router_strategy: della # average # random_init
  blend_experts: true
dtype: float32
out_dtype: bfloat16
tokenizer:
  source: union
chat_template: auto

📜 Merge Log 31B

Merge Method

This model was merged using the KARCHER merge method.

Models Merged

The following models were included in the merge:

📋 31B Yaml

architecture: Gemma4ForConditionalGeneration
models:
  - model: B:/31B/DavidAU--gemma-4-31B-it-Grand-Horror-X-INTENSE-HERETIC-UNCENSORED-Thinking
  - model: B:/31B/MRockatansky--Gemma-4-31B-Storymaxxed3
merge_method: karcher
parameters:  
  max_iter: 30
  tol: 1.0e-9
dtype: float32
out_dtype: bfloat16
tokenizer:
  source: union
chat_template: auto

🧙‍♀️ Channeling Ritual

Recommended Sampling Parameters

temperature: 0.8
top_p: 0.95
top_k: 100
repetition_penalty: 1.12

Downloads last month: 16

Safetensors

Model size

32B params

Tensor type

BF16

Model tree for Naphula/Erebus-RP-Engine-31B

DavidAU/gemma-4-31B-it-Grand-Horror-X-INTENSE-HERETIC-UNCENSORED-Thinking

MRockatansky/Gemma-4-31B-Storymaxxed3

Merge model

this model

Quantizations

2 models

Collections including Naphula/Erebus-RP-Engine-31B

Merge Experiments

Collection

Sorted from oldest (top) to newest (bottom) • 164 items • Updated about 22 hours ago • 4

Favorite Uncensored Drivers

Collection

These models have no refusals and require no jailbreaks. • 77 items • Updated about 22 hours ago • 26

Papers for Naphula/Erebus-RP-Engine-31B

Functionality-Oriented LLM Merging on the Fisher--Rao Manifold

Paper • 2603.04972 • Published Mar 5 • 4

DELLA-Merging: Reducing Interference in Model Merging through Magnitude-Based Sampling

Paper • 2406.11617 • Published Jun 17, 2024 • 12