Instructions to use nvidia/OpenReasoning-Nemotron-14B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nvidia/OpenReasoning-Nemotron-14B with Transformers:

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

pipe = pipeline("text-generation", model="nvidia/OpenReasoning-Nemotron-14B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nvidia/OpenReasoning-Nemotron-14B")
model = AutoModelForCausalLM.from_pretrained("nvidia/OpenReasoning-Nemotron-14B")
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
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use nvidia/OpenReasoning-Nemotron-14B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nvidia/OpenReasoning-Nemotron-14B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/OpenReasoning-Nemotron-14B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nvidia/OpenReasoning-Nemotron-14B

SGLang

How to use nvidia/OpenReasoning-Nemotron-14B 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 "nvidia/OpenReasoning-Nemotron-14B" \
    --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": "nvidia/OpenReasoning-Nemotron-14B",
		"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 "nvidia/OpenReasoning-Nemotron-14B" \
        --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": "nvidia/OpenReasoning-Nemotron-14B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nvidia/OpenReasoning-Nemotron-14B with Docker Model Runner:
```
docker model run hf.co/nvidia/OpenReasoning-Nemotron-14B
```

Is there a way to enable genselect or a tutorial for it?

by brownb2 - opened Jul 21, 2025

Discussion

brownb2

Jul 21, 2025

As title, a tutorial or readme update would be nice...

igitman

NVIDIA org Jul 21, 2025

You can find the commands for how to run it in here https://nvidia.github.io/NeMo-Skills/releases/openreasoning/evaluation/#run-evaluation. We will be adding more details and documentation soon. A basic idea is very simple, you add N generation summaries back into the prompt https://github.com/NVIDIA/NeMo-Skills/blob/main/nemo_skills/prompt/config/openmath/genselect.yaml and then the model produces an index of the best solution. When you have more generations than can fit into the context, you can do either a random subsampling or a tournament between multiple generations.

You can find some more details in these papers https://arxiv.org/abs/2504.16891, https://openreview.net/forum?id=8LhnmNmUDb.

Let us know if you have further questions.

igitman

NVIDIA org Aug 1, 2025

There is also a little standalone script you can use https://huggingface.co/nvidia/OpenReasoning-Nemotron-14B/blob/main/genselect_hf.py

igitman changed discussion status to closed Aug 1, 2025

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