Instructions to use Vortex5/Lunar-Abyss-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Vortex5/Lunar-Abyss-12B with Transformers:

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

pipe = pipeline("text-generation", model="Vortex5/Lunar-Abyss-12B")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Vortex5/Lunar-Abyss-12B")
model = AutoModelForCausalLM.from_pretrained("Vortex5/Lunar-Abyss-12B")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use Vortex5/Lunar-Abyss-12B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Vortex5/Lunar-Abyss-12B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Vortex5/Lunar-Abyss-12B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Vortex5/Lunar-Abyss-12B

SGLang

How to use Vortex5/Lunar-Abyss-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 "Vortex5/Lunar-Abyss-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": "Vortex5/Lunar-Abyss-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 "Vortex5/Lunar-Abyss-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": "Vortex5/Lunar-Abyss-12B",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Vortex5/Lunar-Abyss-12B with Docker Model Runner:
```
docker model run hf.co/Vortex5/Lunar-Abyss-12B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

🌘 Lunar-Abyss-12B

Born where moonlight touches the deep — thought meets desire, and reason dreams.

🌑 Overview

Lunar-Abyss-12B was made to combine the coherency and stability of LunaMaid-12B with the evocative prose and edgy flair of Abyssal-Seraph-12B.

⚖️ Merge Method — DELLA

🧩 Base: Vortex5/MegaMoon-Karcher-12B
💎 Inputs: Vortex5/LunaMaid-12B + Vortex5/Abyssal-Seraph-12B

Configuration

models:
  - model: Vortex5/LunaMaid-12B
    parameters:
      weight:
        - filter: self_attn
          value: [0.35, 0.4, 0.6, 0.8, 1.0, 0.9, 0.6, 0.3]
        - filter: mlp
          value: [0.20, 0.25, 0.35, 0.45, 0.45, 0.40, 0.30, 0.20]
        - value: [0.25, 0.3, 0.35, 0.4, 0.4, 0.35, 0.3, 0.25]
      density: 0.55
      epsilon: 0.3

  - model: Vortex5/Abyssal-Seraph-12B
    parameters:
      weight:
        - filter: mlp
          value: [0.3, 0.5, 0.8, 1.0, 1.0, 0.9, 0.7, 0.4]
        - value: [0.2, 0.3, 0.4, 0.5, 0.5, 0.4, 0.3, 0.2]
      density: 0.5
      epsilon: 0.4
merge_method: della
base_model: Vortex5/MegaMoon-Karcher-12B
parameters:
  lambda: 1.0
  normalize: true
dtype: bfloat16
tokenizer:
  source: Vortex5/Abyssal-Seraph-12B

🌌 Essence of the Merge

Like moonlight reflecting on dark water, Lunar-Abyss carries both clarity and depth.
It thinks with the calm focus of LunaMaid yet speaks with the emotional pulse of Abyssal-Seraph.
Every response flows with a quiet duality — logic beneath, creativity above — neither overpowering the other.

For fans of expressive writing and immersive roleplay, it offers a tone that’s reflective, and mysterious.

🎭 Roleplay & Creative Focus

Designed for narrative storytelling, introspective dialogue, and emotion-driven writing.

🌒 Acknowledgements 🌘

⚙️ mradermacher — static / imatrix quantization
🜛 DeathGodlike — EXL3 quants
🩶 All original model authors and contributors whose work made this model possible.

🧾 Models Merged

Models merged in this creation:

Downloads last month: 7

Safetensors

Model size

12B params

Tensor type

BF16

Model tree for Vortex5/Lunar-Abyss-12B

Vortex5/Abyssal-Seraph-12B

Vortex5/LunaMaid-12B

Vortex5/MegaMoon-Karcher-12B

Merge model

this model

Merges

2 models

Quantizations

4 models