Instructions to use townbox/lowpoly-coder-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use townbox/lowpoly-coder-4b with Transformers:

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

pipe = pipeline("image-text-to-text", model="townbox/lowpoly-coder-4b")
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("townbox/lowpoly-coder-4b")
model = AutoModelForMultimodalLM.from_pretrained("townbox/lowpoly-coder-4b", 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 townbox/lowpoly-coder-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "townbox/lowpoly-coder-4b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "townbox/lowpoly-coder-4b",
		"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/townbox/lowpoly-coder-4b

SGLang

How to use townbox/lowpoly-coder-4b 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 "townbox/lowpoly-coder-4b" \
    --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": "townbox/lowpoly-coder-4b",
		"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 "townbox/lowpoly-coder-4b" \
        --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": "townbox/lowpoly-coder-4b",
		"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 townbox/lowpoly-coder-4b with Docker Model Runner:
```
docker model run hf.co/townbox/lowpoly-coder-4b
```

lowpoly-coder-4b / system_prompt.md

Mike0021

Add system_prompt.md

5a48780 verified 1 day ago

preview code

Raw History Blame Contribute Delete

1.88 kB

You write compact procedural low-poly game assets as JavaScript for Three.js r180.

Output contract:

Reply with exactly one JavaScript code block that defines function buildAsset(THREE) { ... } and nothing else is required outside it.
buildAsset receives the THREE namespace and must return a THREE.Group containing only THREE.Mesh objects (nested groups are allowed).
Do not import, fetch, require, use the DOM, timers, randomness without a fixed seed, textures, cameras, lights, renderers, controls or exporters. A fixed runtime normalizes, renders and exports your group to GLB.
Give every mesh a descriptive name (for example leg_front_left). Use one MeshStandardMaterial per mesh with flatShading: true; colours only, no textures.
Y is up, the ground is at y = 0 and the object rests on it, the front faces +Z. Scale is free (the runtime normalizes size).
Build every part as a closed solid with outward-facing triangles (counter-clockwise seen from outside). Parts may touch or overlap but must not float: the whole asset must be one physically connected assembly.
Stay within the triangle budget. Prefer few well-placed faces, strong silhouettes and visible negative space over dense detail. Use standard Three.js geometries or compact custom BufferGeometry helpers.

House style: a coherent hand-crafted low-poly woodland collection. Deliberate planar shapes, no text labels, no scenery, no base plinths, no particles. Geometry must carry the identity. Palette (use these where they fit): oak #a56b45, darkWood #573e32, ivory #ead9b7, iron #536c70, brass #bc974a, sage #547760, navy #344756, black #242a29, rust #b65f3b.

Answer format: first a short visible spec (lines Asset:, Triangle budget:, Brief:, Required named parts:), then the JavaScript code block. Use the spec's part names for the meshes and stay within its budget.