Instructions to use DoppelReflEx/MiniusLight-24B-v2.1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use DoppelReflEx/MiniusLight-24B-v2.1 with Transformers:

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

pipe = pipeline("text-generation", model="DoppelReflEx/MiniusLight-24B-v2.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("DoppelReflEx/MiniusLight-24B-v2.1")
model = AutoModelForCausalLM.from_pretrained("DoppelReflEx/MiniusLight-24B-v2.1")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use DoppelReflEx/MiniusLight-24B-v2.1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "DoppelReflEx/MiniusLight-24B-v2.1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "DoppelReflEx/MiniusLight-24B-v2.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/DoppelReflEx/MiniusLight-24B-v2.1

SGLang

How to use DoppelReflEx/MiniusLight-24B-v2.1 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 "DoppelReflEx/MiniusLight-24B-v2.1" \
    --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": "DoppelReflEx/MiniusLight-24B-v2.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "DoppelReflEx/MiniusLight-24B-v2.1" \
        --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": "DoppelReflEx/MiniusLight-24B-v2.1",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use DoppelReflEx/MiniusLight-24B-v2.1 with Docker Model Runner:
```
docker model run hf.co/DoppelReflEx/MiniusLight-24B-v2.1
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

MiniusLight-24B-v2.1

12B - 24B-v1 - 24B-v1.01 - 24B-v2 - 24B-v2.1

Origin Content (Click Here)

What is this?

A merge of most uncensored model TroyDoesAI/BlackSheep-24B and recipe of MiniusLight-24B: TheDrummer/Cydonia-24B-v2 and PocketDoc/Dans-PersonalityEngine-V1.2.0-24b.

Another version of v2, but far better than it. Vivid writing styles, and talk back to me, sometimes hard to control it. (Maybe just because my character card)

Best model of the series (for me). :)

PS: Highest NatInt for 24B model in UGI leaderboard (1st May 2025)

GGUF (Thank mradermacher and his team so much (nicoboss too))

Static - iMatrix

Other information

Chat Template? ChatML, of course! Mistral V7 if you want the model smarter.

Merge Method

Detail YAML Config

  {
    models:
     - model: TroyDoesAI/BlackSheep-24B
       parameters:
         density: 0.9
         weight: 1
     - model: TheDrummer/Cydonia-24B-v2
       parameters:
         density: 0.6
         weight: 0.8
     - model: PocketDoc/Dans-PersonalityEngine-V1.2.0-24b
       parameters:
         density: 0.8
         weight: 0.6
    merge_method: dare_ties
    base_model: TroyDoesAI/BlackSheep-24B
    tokenizer_source: base
    parameters:
      rescale: true
    dtype: bfloat16
  }