Instructions to use xsdfasfgsa/Decision-King with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use xsdfasfgsa/Decision-King with Transformers:

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

pipe = pipeline("image-text-to-text", model="xsdfasfgsa/Decision-King")
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("xsdfasfgsa/Decision-King")
model = AutoModelForMultimodalLM.from_pretrained("xsdfasfgsa/Decision-King", 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 xsdfasfgsa/Decision-King with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "xsdfasfgsa/Decision-King"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "xsdfasfgsa/Decision-King",
		"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/xsdfasfgsa/Decision-King

SGLang

How to use xsdfasfgsa/Decision-King 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 "xsdfasfgsa/Decision-King" \
    --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": "xsdfasfgsa/Decision-King",
		"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 "xsdfasfgsa/Decision-King" \
        --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": "xsdfasfgsa/Decision-King",
		"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"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use xsdfasfgsa/Decision-King with Docker Model Runner:
```
docker model run hf.co/xsdfasfgsa/Decision-King
```

Decision-King

Decision-King (model repository) was initialized from Qwen/Qwen3.8-27B at commit 1d4bf0f2ff6012fd82039f2fa52739d0dd7c60c0.

We did not train on JevBench; its public split was used for evaluation and tuning.

Requires a BF16-capable CUDA GPU with sufficient VRAM (80 GB recommended).

Download and install

python -m pip install --upgrade huggingface_hub
hf download xsdfasfgsa/Decision-King --local-dir Decision-King
cd Decision-King
python -m pip install -r requirements.txt

Run inference

CUDA_VISIBLE_DEVICES=0 python -B inference/infer.py --model-path . --request example_request.json

example_request.json format (state is your context; questions contains your decisions):

{
  "state": {
    "record": "Independent plumbing example, not a benchmark item.",
    "requested_units": 7,
    "available_units": 12
  },
  "questions": {
    "enough": {"type": "noul", "instructions": "Are available units at least requested units?"},
    "route": {"type": "choice", "instructions": "Which team owns a stock-count request?", "criteria": {"inventory": "Stock counting", "payments": "Payment processing", "facilities": "Building maintenance"}},
    "capacity": {"type": "score", "instructions": "Grade the remaining available units after fulfilling the request.", "criteria": ["Fewer than zero remain.", "Between zero and four inclusive remain.", "Five or more remain."]}
  }
}

Replace example_request.json with your own request file using the same format.

JevBench raw input

Raw JevBench JSON/JSONL records are accepted directly; no separate conversion step is needed:

cat /path/to/jevbench/datasets/public/{easy,original,hard}.jsonl > jevbench.jsonl
CUDA_VISIBLE_DEVICES=0 python -B inference/infer.py --model-path . --request jevbench.jsonl --jsonl > predictions.jsonl

The adapter preserves the original label order and never passes expected or benchmark metadata to the model. Raw-task responses also include task_id and probs. Accuracy is computed separately with the JevBench scorer.

HTTP server (optional)

CUDA_VISIBLE_DEVICES=0 python -B inference/serve.py --model-path . --host 127.0.0.1 --port 8000

From another terminal in the repository directory:

curl -s http://127.0.0.1:8000/v1/systemone -H 'Content-Type: application/json' --data-binary @example_request.json

Downloads last month: 33

Safetensors

Model size

27B params

Tensor type

BF16