Image-Text-to-Text
Transformers
Safetensors
qwen3_5
decision-model
typed-decisions
multimodal
conversational
Instructions to use inclusionAI/SingVL_S1_4B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use inclusionAI/SingVL_S1_4B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="inclusionAI/SingVL_S1_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("inclusionAI/SingVL_S1_4B") model = AutoModelForMultimodalLM.from_pretrained("inclusionAI/SingVL_S1_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 inclusionAI/SingVL_S1_4B with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "inclusionAI/SingVL_S1_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": "inclusionAI/SingVL_S1_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/inclusionAI/SingVL_S1_4B
- SGLang
How to use inclusionAI/SingVL_S1_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 "inclusionAI/SingVL_S1_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": "inclusionAI/SingVL_S1_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 "inclusionAI/SingVL_S1_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": "inclusionAI/SingVL_S1_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 inclusionAI/SingVL_S1_4B with Docker Model Runner:
docker model run hf.co/inclusionAI/SingVL_S1_4B
Download inference/engine.py from inclusionAI/SingVL_S1_4B: direct link, hf CLI and curl.
- Browser
- Download file 2.18 kB
-
https://huggingface.co/inclusionAI/SingVL_S1_4B/resolve/main/inference/engine.py
- Command line
-
hf download hf://inclusionAI/SingVL_S1_4B/inference/engine.py
-
curl -L -o engine.py https://huggingface.co/inclusionAI/SingVL_S1_4B/resolve/main/inference/engine.py
2.18 kB
| """Decision Index native-engine adapter using the released Transformers readout.""" | |
| from io import BytesIO | |
| from pathlib import Path | |
| from PIL import Image | |
| from decision_index.engines.base import Engine, Unsupported as KitUnsupported | |
| from decision_index.vision.images import load | |
| from .model import SingVLS1, Unsupported, PUBLIC_IMAGE_MAX_PIXELS | |
| class D3Engine(Engine): | |
| name = "sing-vl-s1" | |
| latency = "One native Transformers prediction including input preparation; model loading excluded." | |
| def __init__(self, model, device="cuda:0", image_root=None, | |
| image_max_pixels=PUBLIC_IMAGE_MAX_PIXELS, **options): | |
| super().__init__(**options) | |
| self.image_root = image_root | |
| self.native = SingVLS1(model, device=device, image_max_pixels=image_max_pixels) | |
| self.provenance = { | |
| "kind": "native-transformers", "model": str(Path(model).resolve()), | |
| "max_pixels": image_max_pixels, | |
| "image_resize": "native Sing VL S1 processor: preserve aspect ratio, patch-grid rounding, RGB bicubic resize, 1.6 MP cap", | |
| "attention": "prefix-bidirectional full attention, causal GDN, no KV cache", | |
| "max_images": 4, "text_max_tokens": 65536, "vision_max_tokens": 10000, | |
| } | |
| def __call__(self, state, questions, images=(), prompt=None): | |
| decoded = [] | |
| for ref in images: | |
| with Image.open(BytesIO(load(ref, self.image_root))) as image: | |
| decoded.append(image.copy()) | |
| try: | |
| response, details = self.native.predict(state, questions, decoded) | |
| except Unsupported as exc: | |
| raise KitUnsupported(str(exc)) from exc | |
| grids = next(iter(details.values())).get("image_grid_thw") | |
| factor = self.native.image_factor | |
| raw = {"details": details, "prompt_metadata": prompt, | |
| "image_pixels": [int(t*h*w*factor*factor/self.native.merge_size**2) for t,h,w in grids or []], | |
| "image_tokens": next(iter(details.values())).get("image_tokens", 0)} | |
| return response, raw | |
| def runtime(self): | |
| return self.native.runtime() | |
| def close(self): | |
| del self.native | |