Instructions to use chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview with Transformers:

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

pipe = pipeline("image-text-to-text", model="chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview")
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)

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview")
model = AutoModelForImageTextToText.from_pretrained("chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview")
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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview",
		"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/chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview

SGLang

How to use chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview 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 "chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview" \
    --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": "chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview",
		"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 "chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview" \
        --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": "chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview",
		"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 chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview with Docker Model Runner:
```
docker model run hf.co/chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview
```

EchoVLM (paper implementation)

Official PyTorch implementation of the model described in
"EchoVLM: Dynamic Mixture-of-Experts Vision-Language Model for Universal Ultrasound Intelligence".

🤖 Model Details

Item	Value
Paper	arXiv:2509.14977
Authors	Chaoyin She¹, Ruifang Lu²
Code	GitHub repo
Model Hub	Hugging Face

🔄 Updates

Coming soon: V2 with Chain-of-Thought reasoning and reinforcement learning enhancements—full training & inference code plus benchmark test-set will be fully open-sourced.
Dec 1, 2025: To better promote development in this field, we've open-sourced our latest instruction fine-tuned model based on Lingshu-7B. Essentially built on Qwen2.5VL, it enjoys a better ecosystem—for example, it can seamlessly leverage vLLM for accelerated inference. Released model weights on Hugging Face.
Sep 21, 2025: The full, uncleaned model codebase is now open-sourced on GitHub!
Sep 19, 2025: Released model weights on Hugging Face.
Sep 17, 2025: Paper published on arXiv.

🚀 Quick Start

Reference Qwen2.5-VL-7B-Instruct

📌 Citation

If you use this model or code in your research, please cite:

@misc{she2025echovlmdynamicmixtureofexpertsvisionlanguage,
      title={EchoVLM: Dynamic Mixture-of-Experts Vision-Language Model for Universal Ultrasound Intelligence}, 
      author={Chaoyin She and Ruifang Lu and Lida Chen and Wei Wang and Qinghua Huang},
      year={2025},
      eprint={2509.14977},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2509.14977}, 
}

Downloads last month: 186

Safetensors

Model size

8B params

Tensor type

BF16

Model tree for chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview

Base model

lingshu-medical-mllm/Lingshu-7B

Finetuned

(3)

this model

Paper for chaoyinshe/EchoVLM_V2_lingshu_base_7b_instruct_preview

EchoVLM: Dynamic Mixture-of-Experts Vision-Language Model for Universal Ultrasound Intelligence

Paper • 2509.14977 • Published Sep 18, 2025 • 5