Instructions to use dphn/Dolphin-Mistral-24B-Venice-Edition with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use dphn/Dolphin-Mistral-24B-Venice-Edition with Transformers:

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

pipe = pipeline("text-generation", model="dphn/Dolphin-Mistral-24B-Venice-Edition")
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, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("dphn/Dolphin-Mistral-24B-Venice-Edition")
model = AutoModelForMultimodalLM.from_pretrained("dphn/Dolphin-Mistral-24B-Venice-Edition", 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=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use dphn/Dolphin-Mistral-24B-Venice-Edition with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "dphn/Dolphin-Mistral-24B-Venice-Edition"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "dphn/Dolphin-Mistral-24B-Venice-Edition",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/dphn/Dolphin-Mistral-24B-Venice-Edition

SGLang

How to use dphn/Dolphin-Mistral-24B-Venice-Edition 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 "dphn/Dolphin-Mistral-24B-Venice-Edition" \
    --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": "dphn/Dolphin-Mistral-24B-Venice-Edition",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "dphn/Dolphin-Mistral-24B-Venice-Edition" \
        --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": "dphn/Dolphin-Mistral-24B-Venice-Edition",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use dphn/Dolphin-Mistral-24B-Venice-Edition with Docker Model Runner:
```
docker model run hf.co/dphn/Dolphin-Mistral-24B-Venice-Edition
```

Is this a new model?

by sonicrules1234 - opened Jun 18, 2025

Discussion

sonicrules1234

Jun 18, 2025

I know that there was a dolphin venice mistral model before this. What happened to that one, and what is the difference between that one and this one?

dphnAI

Dolphin org Jun 19, 2025

This is an updated version based on feedback we received on v1.
It is the first time we have used RL to remove censorship from a model, and it also includes new datasets aimed at making the model less dry (the 24B is very STEM-oriented).

The older version can still be accessed privately here with this key for anyone who still wants to use it

cognitivecomputations/Dolphin-Mistral-24B-Venice-Edition-Legacy
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

It can be run with

export HF_TOKEN="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"

before your startup command

--token "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" \

in the startup command

Vali

Jun 19, 2025

If it is a new updated version can it be renamed to be more obvious? With V2 in the name or something, it whoud be less confusing.

mradermacher

Jun 19, 2025

Yes, when silently replacing a model in the same repo, there will be endless confusion on which model (when converted into other formats) is which, a situation which sucks especially for end-users - nobody will know what model is the real one.

darkc0de

Jun 20, 2025

Request DOI?

EloyOn

Jul 19, 2025

•

edited Jul 20, 2025

Yeah, it should have v1.1 in the name to avoid confusion. I thought I already downloaded this LLM is turns out it's an updated version.

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