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

Libraries

How to use Cloudflare/clef with Transformers:

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

pipe = pipeline("image-text-to-text", model="Cloudflare/clef")
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("Cloudflare/clef")
model = AutoModelForMultimodalLM.from_pretrained("Cloudflare/clef", 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 Cloudflare/clef with vLLM:

Install from pip and serve model

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

SGLang

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

Demo for this model on Spaces

by multimodalart - opened Oct 1

Discussion

multimodalart

Oct 1

•

edited Oct 1

Hi @areneau 🤗

I'm Apolinario, from the open-source team at Hugging Face. Congrats and thanks for open-sourcing Cloudflare/clef on the Hub! We were excited about this work and built with an agent an interactive demo app of it on Hugging Face Spaces, running on a free ZeroGPU infrastructure.

Here's a link to the demo: https://huggingface.co/spaces/hugging-apps/clef-decision-model

We would love to transfer this demo to you or your organization. Would you like this demo to live under your own account or organization? If so just let me know here which username to transfer to, and we'll transfer the Space over to you, we hope it can give your work more visibility, discoverability and allows folks to try it out.

(If you have any questions or just want to chat more about this, you can find me on Twitter, LinkedIn or apolinario @ huggingface.co)

Cheers,

Poli

MilestoneAI

about 1 month ago

fails on this:

"Hi team, hope you're all having a peaceful weekend. There is absolutely no rush on this, so please take your time! While reviewing our weekly GDPR compliance logs internally, we noticed a minor quirk: it looks like hashed user identifiers from our EU tenant might occasionally pass unmasked through the secondary webhook endpoints when syncing with our backup compliance vault. The system is fully operational and there's zero downtime, so it's purely a data governance detail. Just wanted to flag it whenever someone has a spare moment next week. Thanks as always!"

The decision was:
{
"model": "clef",
"answers": {
"urgent": {
"type": "noul",
"noul": 0.0087
},
"team": {
"type": "choice",
"choice": "technical",
"confidence": 0.9391,
"probabilities": {
"billing": 0.0314,
"technical": 0.9391,
"sales": 0.0295
}
},
"severity": {
"type": "score",
"score": 0.7365,
"confidence": 0.6226,
"legend": {
"0": "No impact",
"1": "Minor",
"2": "Major",
"3": "Critical"
},
"probabilities": {
"0": 0.3274,
"1": 0.6226,
"2": 0.0361,
"3": 0.0139
}
}
},
"usage": {
"input_tokens": 445,
"output_tokens": 0
}
}

the wording of the prompt makes the model feel this is a minor issue, but this is a Critical / P1 Security & Compliance Incident. The mention of unmasked user identifiers crossing European tenants without proper governance triggers potential GDPR violations and high legal/regulatory exposure for an enterprise client.

laserbeans

about 1 month ago

the wording of the prompt makes the model feel this is a minor issue, but this is a Critical / P1 Security & Compliance Incident. The mention of unmasked user identifiers crossing European tenants without proper governance triggers potential GDPR violations and high legal/regulatory exposure for an enterprise client.

Need GDPRBench

MilestoneAI

about 1 month ago

the wording of the prompt makes the model feel this is a minor issue, but this is a Critical / P1 Security & Compliance Incident. The mention of unmasked user identifiers crossing European tenants without proper governance triggers potential GDPR violations and high legal/regulatory exposure for an enterprise client.

Need GDPRBench

Not really. unmasked user identifiers should be read as a red flag, even when disguised in a casual wording.

Keithdickey207

29 days ago

Ways to Tune It for Clef
Add Contextual Anchors in urgent: Since noul returns a true/false probability, making the instruction more specific helps Clef weigh edge cases (e.g., "Does this request require a response within 1 hour?" rather than just "Is this urgent?").

Sharpen the choice Criteria: Clef uses the description strings under criteria to map semantic intent. Making those definitions slightly more mutually exclusive reduces probability bleed between categories.

Optimized Schema Example for Clef
JSON
{
"urgent": {
"type": "noul",
"instructions": "Does this request indicate an immediate blockage, outage, or financial emergency requiring intervention within the hour?"
},
"team": {
"type": "choice",
"instructions": "Which department has the exact domain expertise required to resolve this request?",
"criteria": {
"billing": "Inquiries strictly concerning credit cards, payment failures, past invoices, or refund requests.",
"technical": "Reports of software bugs, API errors, system downtime, or unexpected application behavior.",
"sales": "Inquiries regarding custom pricing tiers, upgrade paths, enterprise features, or new subscriptions."
}
},
"severity": {
"type": "score",
"instructions": "What is the degree of operational disruption experienced by the user?",
"criteria": [
"No impact",
"Minor",
"Major",
"Critical"
]
}
}

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