Instructions to use deepseek-ai/deepseek-vl-7b-chat with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use deepseek-ai/deepseek-vl-7b-chat with Transformers:

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

pipe = pipeline("image-text-to-text", model="deepseek-ai/deepseek-vl-7b-chat")

# Load model directly
from transformers import MultiModalityCausalLM
model = MultiModalityCausalLM.from_pretrained("deepseek-ai/deepseek-vl-7b-chat", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use deepseek-ai/deepseek-vl-7b-chat with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "deepseek-ai/deepseek-vl-7b-chat"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deepseek-ai/deepseek-vl-7b-chat",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/deepseek-ai/deepseek-vl-7b-chat

SGLang

How to use deepseek-ai/deepseek-vl-7b-chat 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 "deepseek-ai/deepseek-vl-7b-chat" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deepseek-ai/deepseek-vl-7b-chat",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "deepseek-ai/deepseek-vl-7b-chat" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deepseek-ai/deepseek-vl-7b-chat",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use deepseek-ai/deepseek-vl-7b-chat with Docker Model Runner:
```
docker model run hf.co/deepseek-ai/deepseek-vl-7b-chat
```

Model is not inferencing on multiple images; is this the right template?

by ltbd78 - opened Mar 15, 2024

Discussion

ltbd78

Mar 15, 2024

conversation = [
    {
        "role": "User",
        "content": "Compare and contrast <image_placeholder> and <image_placeholder>.",
        "images": ["./data/1.png", "./data/2.jpg"]
    },
    {
        "role": "Assistant",
        "content": ""
    }
]

ltbd78

Mar 15, 2024

•

edited Mar 15, 2024

UPDATE: it worked for a different set of images and prompt "Describe <image_placeholder>. Then describe <image_placeholder>."
Though, I would like to clarify: how is order determined? Is it sequential?

doubility123

DeepSeek org Mar 18, 2024

•

edited Mar 18, 2024

Yes, for the multi-image inputs, your prompt is correct. Additionally, these images are sequential. The first <image_placeholder> corresponds to "./data/1.png," while the second <image_placeholder> corresponds to "./data/2.png."

SinanAkkoyun

Apr 17, 2024

This comment has been hidden

SinanAkkoyun

Apr 17, 2024

@doubility123 For me, multi image input does not work at all. Was deepseek VL trained on multi image-text pairs?

For our model, we need to train on multi image-text pairs, does the architecture support that?

Thanks

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment