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

Libraries

How to use windprak/open_steuerllm with Transformers:

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

pipe = pipeline("text-generation", model="windprak/open_steuerllm")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("windprak/open_steuerllm")
model = AutoModelForCausalLM.from_pretrained("windprak/open_steuerllm")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use windprak/open_steuerllm with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/windprak/open_steuerllm

SGLang

How to use windprak/open_steuerllm 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 "windprak/open_steuerllm" \
    --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": "windprak/open_steuerllm",
		"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 "windprak/open_steuerllm" \
        --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": "windprak/open_steuerllm",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use windprak/open_steuerllm with Docker Model Runner:
```
docker model run hf.co/windprak/open_steuerllm
```

Benchmark Submission: HTTP 429 on all IPs (Bug?)

by bleron112 - opened Feb 13

Discussion

bleron112

Feb 13

Hi,

I've been trying to submit predictions to the SteuerEx Benchmark
(https://steuerllm.i5.ai.fau.de/benchmark) for Claude Opus 4,
but every attempt returns:

"You have already submitted 1 time(s). Maximum 1 submission(s) per IP allowed."

I've tried 4 completely different networks/IPs (home WiFi, mobile hotspot,
VPN Zurich, VPN Tirana) — all return the same 429 error. This suggests
the server might be checking something other than the IP, or there's
a bug in the submission system.

My predictions.json is validated (115 answers, all IDs 1001-1115,
UTF-8, no empty entries).

Could you look into this? I'd love to test multiple models on the benchmark.

Also — is there any way to get access to the reference statements
for local evaluation?

Thanks for the great benchmark!

windprak

Owner Feb 13

It was a configuration issue on our side. Can you try:

curl -s -X POST https://steuerllm.i5.ai.fau.de/benchmark/submit -F "model_name=TestModel" -F "key=XX" -F "file=@/home/user/test_submission.json" 2>&1

{"message":"Submission received and queued for evaluation","queue_position":1,"status_url":"/status/4c112a8e38783049","submission_id":"4c112a8e38783049","success":true}

curl -s https://steuerllm.i5.ai.fau.de/benchmark/status/ID 2>&1 | python3 -m json.tool
{
"model_name": "TestModel",
"progress": 42,
"queue_position": 1,
"queue_size": 0,
"status": "evaluating",
"timestamp": "2026-02-13T10:18:57.815690"
}

The gold answers will be released when the service is taken down. For now we want to prevent training polution.

bleron112

Feb 14

thanks

bleron112 changed discussion status to closed Feb 14

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