Instructions to use utter-project/EuroLLM-22B-Instruct-2512 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use utter-project/EuroLLM-22B-Instruct-2512 with Transformers:

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

pipe = pipeline("text-generation", model="utter-project/EuroLLM-22B-Instruct-2512")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForMultimodalLM

tokenizer = AutoTokenizer.from_pretrained("utter-project/EuroLLM-22B-Instruct-2512")
model = AutoModelForMultimodalLM.from_pretrained("utter-project/EuroLLM-22B-Instruct-2512")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use utter-project/EuroLLM-22B-Instruct-2512 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "utter-project/EuroLLM-22B-Instruct-2512"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "utter-project/EuroLLM-22B-Instruct-2512",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/utter-project/EuroLLM-22B-Instruct-2512

SGLang

How to use utter-project/EuroLLM-22B-Instruct-2512 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 "utter-project/EuroLLM-22B-Instruct-2512" \
    --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": "utter-project/EuroLLM-22B-Instruct-2512",
		"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 "utter-project/EuroLLM-22B-Instruct-2512" \
        --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": "utter-project/EuroLLM-22B-Instruct-2512",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use utter-project/EuroLLM-22B-Instruct-2512 with Docker Model Runner:
```
docker model run hf.co/utter-project/EuroLLM-22B-Instruct-2512
```

Used for translation of industrial PLC docs — thanks

by JonathanVanpeteghem - opened 27 days ago

Discussion

JonathanVanpeteghem

27 days ago

Hi EuroLLM team,

Quick thank-you note. We've adopted EuroLLM-22B-Instruct-2512 (Q8_0) as the default local translation model in a production document-translation tool. We use it across Dutch, English, French, German, Italian, Polish, Portuguese and Spanish; Dutch source is currently where the workload sits.

Operating point: we send 8 short technical inputs per call as a numbered list — a mix of alarm strings, HMI screen labels, parameter labels, operator-instruction prose, table cells, and document headings. ~280 prompt-tokens / ~165 completion-tokens per call, ~50–60 s end-to-end on the batched path. 4-way client-side parallelism against a single Ollama backend; we're evaluating an NVIDIA DGX Spark as the host going forward.

Industrial-documentation translation is a nasty corner of the space — alarm strings mixed with PLC tags, polysemous Dutch verbs, brand pass-throughs, ALL-CAPS labels — and EuroLLM-22B handles it well. Citation added to our project's NOTICES.md per the model card.

Thanks for releasing it under Apache 2.0 and for the work it represents. Hope the project continues.

Jonathan

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