Instructions to use Naphula/GhostFace-24B-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Naphula/GhostFace-24B-v1 with Transformers:

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

pipe = pipeline("text-generation", model="Naphula/GhostFace-24B-v1")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Naphula/GhostFace-24B-v1")
model = AutoModelForCausalLM.from_pretrained("Naphula/GhostFace-24B-v1")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Naphula/GhostFace-24B-v1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Naphula/GhostFace-24B-v1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Naphula/GhostFace-24B-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Naphula/GhostFace-24B-v1

SGLang

How to use Naphula/GhostFace-24B-v1 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/GhostFace-24B-v1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Naphula/GhostFace-24B-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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/GhostFace-24B-v1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Naphula/GhostFace-24B-v1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Naphula/GhostFace-24B-v1 with Docker Model Runner:
```
docker model run hf.co/Naphula/GhostFace-24B-v1
```
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 Mistral Tekken chat template.

👻 GhostFace 24B v1

Merge Method

This model was merged using the scream method. This custom method combines the stability of model_stock with the novelty of della and sce.

This is a random experiment that seeks to unite several of the best Mistral 24B merges and finetunes. There are some refusals, however the model seems quite intelligent so is being released as is.

I don't feel like ablating it right now but maybe later since it works pretty well with jailbreaks.

architecture: MistralForCausalLM
merge_method: scream # Similarity-Consensus Resolved Enhanced Adaptive Merging
base_model: B:\24B\!models--mistralai--Magistral-Small-2509
models:
  - model: B:\24B\!BeaverAI_Fallen-Mistral-Small-3.1-24B-v1e_textonly
    parameters:
      weight: 0.1
  - model: B:\24B\!models--Casual-Autopsy--Maginum-Cydoms-24B
    parameters:
      weight: 0.1
  - model: B:\24B\!models--DarkArtsForge--Asmodeus-24B-v1
    parameters:
      weight: 0.1
  - model: B:\24B\!models--DarkArtsForge--Magistaroth-24B-v1
    parameters:
      weight: 0.1
  - model: B:\24B\!models--FlareRebellion--WeirdCompound-v1.7-24b
    parameters:
      weight: 0.1
  - model: B:\24B\!models--Gryphe--Tiamat-24B-Magistral
    parameters:
      weight: 0.1
  - model: B:\24B\!models--Naphula--Slimaki-24B-v1
    parameters:
      weight: 0.1
  - model: B:\24B\!models--ReadyArt--4.2.0-Broken-Tutu-24b
    parameters:
      weight: 0.1
  - model: B:\24B\!models--sophosympatheia--Magistry-24B-v1.0
    parameters:
      weight: 0.1
  - model: B:\24B\!models--TheDrummer--Cydonia-24B-v4.3
    parameters:
      weight: 0.1
  - model: B:\24B\!models--TheDrummer--Magidonia-24B-v4.3
    parameters:
      weight: 0.1
  - model: B:\24B\!models--TheDrummer--Precog-24B-v1
    parameters:
      weight: 0.1
  - model: B:\24B\!models--zerofata--MS3.2-PaintedFantasy-v2-24B
    parameters:
      weight: 0.1
  - model: B:\24B\!models--zerofata--MS3.2-PaintedFantasy-v3-24B
    parameters:
      weight: 0.1
  - model: B:\24B\!models--zerofata--MS3.2-PaintedFantasy-v4.1-24B
    parameters:
      weight: 0.1
  - model: B:\24B\!models--Naphula--Goetia-24B-v1.3
    parameters:
      weight: 0.1
parameters:
  stock_weight: 0.4          # Weight for model_stock component
  della_novelty_weight: 0.3  # Weight for DELLA novelty
  sce_novelty_weight: 0.3    # Weight for SCE novelty
  density: 0.9               # DELLA density parameter
  epsilon: 0.05              # DELLA epsilon parameter
  select_topk: 0.5           # SCE top-k selection
  filter_wise: false         # Model Stock filter-wise calculation
  int8_mask: false           # Use int8 masks for memory efficiency
dtype: float32
out_dtype: bfloat16
tokenizer:
  source: union
# chat_template: auto
name: 👻 GhostFace-24B-v1