Instructions to use WasamiKirua/NightShade-Violet-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use WasamiKirua/NightShade-Violet-12B with Transformers:

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

pipe = pipeline("text-generation", model="WasamiKirua/NightShade-Violet-12B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("WasamiKirua/NightShade-Violet-12B")
model = AutoModelForCausalLM.from_pretrained("WasamiKirua/NightShade-Violet-12B", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use WasamiKirua/NightShade-Violet-12B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "WasamiKirua/NightShade-Violet-12B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WasamiKirua/NightShade-Violet-12B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/WasamiKirua/NightShade-Violet-12B

SGLang

How to use WasamiKirua/NightShade-Violet-12B 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 "WasamiKirua/NightShade-Violet-12B" \
    --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": "WasamiKirua/NightShade-Violet-12B",
		"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 "WasamiKirua/NightShade-Violet-12B" \
        --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": "WasamiKirua/NightShade-Violet-12B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use WasamiKirua/NightShade-Violet-12B with Docker Model Runner:
```
docker model run hf.co/WasamiKirua/NightShade-Violet-12B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

💜 Nightshade-Violet-12B (The Emotional Sniper)

Nightshade-Violet-12B is a specialized 12B parameter model engineered for Psychological Precision and Emotional Impact. While its sibling (Nightshade-Lotus) is a storyteller, Violet is an executioner of subtext.

By leveraging the high EQ (Emotional Quotient) of the Violet-Lotus architecture, this model is designed to "read between the lines" and strike the user with surgical, minimalist, and deeply resonant responses.

🧠 The "Surgical" Merge Recipe

This model uses DARE-TIES to prioritize emotional intelligence and brevity, pruning unnecessary prose to ensure every word carries maximum weight.

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: FallenMerick/MN-Violet-Lotus-12B
    parameters:
      weight: 0.45 
      # L'Ancora EQ: Domina la comprensione del sottotesto e dell'emozione.
  - model: Vortex5/Cosmic-Night-12B
    parameters:
      weight: 0.25 
      # Il Silenziatore: Impone la brevità e la struttura secca.
  - model: Vortex5/Crimson-Constellation-12B
    parameters:
      weight: 0.15 
      # Lo Stiletto: Fornisce l'intensità e la capacità di ferire/colpire.
  - model: Vortex5/Moonlit-Mirage-12B
    parameters:
      weight: 0.15 
      # Il Filtro: Aggiunge quel tono cinico e atmosferico "Nightshade".

merge_method: dare_ties
base_model: Vortex5/NoctyxCosma-12B
parameters:
  density: 0.55 
  # Densità media: Abbastanza alta per non perdere l'EQ di Violet, 
  # ma abbastanza bassa da "pulire" la verbosità di base.
  weight: 1.0
dtype: bfloat16
tokenizer_source: base

💪 Key Strengths: The Emotional Edge

Psychological Dominance: The model excels at identifying the user's emotional state and responding in a way that feels unnervingly "aware."

Minimalist Impact: It favors quality over quantity. It won't write paragraphs; it will write three lines that leave you thinking for an hour.

Subtext Recognition: Highly capable of navigating complex relationship dynamics, gaslighting, and indirect communication.

Cold Empathy: It understands emotions perfectly, but uses that understanding to maintain its "Nightshade" persona—calculated, distant, and intense.

🚀 Best Use Cases

Intense 1-on-1 RP: Especially for antagonistic, mysterious, or psychologically complex characters.

Conflict Resolution (or Escalation): Where the drama is internal and verbal rather than physical.

Minimalist Storytelling: For users who prefer a "ping-pong" chat style where the AI reacts sharply to every move.

⚠️ Limitations

Non-Descriptive: If you ask for a description of a room, it might give you a metaphor for your character's loneliness instead. It is not for world-building.

Potentially Abrasive: Its high EQ combined with the "Crimson" weight makes it very good at being mean in a way that feels personal.

High Inference Sensitivity: Small changes in your prompt can drastically change the "vibe" of its strike.

📈 Recommended Inference Settings

To keep the "Sniper" focused and prevent rambling:

Temperature: 0.7 - 0.75 (Lower for more surgical precision).

Min-P: 0.1 (Crucial to filter out low-probability "noise").

Presence Penalty: 0.15 (Keeps the vocabulary fresh without being loquacious).

Frequency Penalty: 0.1

Disclaimer

Nightshade-Violet-12B is a specialized roleplay tool. It is designed to be emotionally provocative and unfiltered. It should be treated as an experimental entity for mature psychological exploration.