NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4

Instructions to use Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 with Transformers:

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

pipe = pipeline("text-generation", model="Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4", trust_remote_code=True, device_map="auto")
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 Settings

vLLM

How to use Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4

SGLang

How to use Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 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 "Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4" \
    --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": "Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4",
		"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 "Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4" \
        --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": "Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 with Docker Model Runner:
```
docker model run hf.co/Loron200/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4
```

NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 / hf_quant_config.json

Loron200

Duplicate from nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4

e27240f 13 days ago

Raw

History Blame Contribute Delete

3.05 kB

	{
	"producer": {
	"name": "modelopt",
	"version": "0.29.0"
	},
	"quantization": {
	"quant_algo": "NVFP4",
	"kv_cache_quant_algo": "FP8",
	"group_size": 16,
	"exclude_modules": [
	"lm_head",
	"backbone.layers.4.mixer.in_proj",
	"backbone.layers.4.mixer.out_proj",
	"backbone.layers.5.mixer.q_proj",
	"backbone.layers.5.mixer.k_proj",
	"backbone.layers.5.mixer.v_proj",
	"backbone.layers.5.mixer.o_proj",
	"backbone.layers.11.mixer.in_proj",
	"backbone.layers.11.mixer.out_proj",
	"backbone.layers.12.mixer.q_proj",
	"backbone.layers.12.mixer.k_proj",
	"backbone.layers.12.mixer.v_proj",
	"backbone.layers.12.mixer.o_proj",
	"backbone.layers.18.mixer.in_proj",
	"backbone.layers.18.mixer.out_proj",
	"backbone.layers.19.mixer.q_proj",
	"backbone.layers.19.mixer.k_proj",
	"backbone.layers.19.mixer.v_proj",
	"backbone.layers.19.mixer.o_proj",
	"backbone.layers.25.mixer.in_proj",
	"backbone.layers.25.mixer.out_proj",
	"backbone.layers.26.mixer.q_proj",
	"backbone.layers.26.mixer.k_proj",
	"backbone.layers.26.mixer.v_proj",
	"backbone.layers.26.mixer.o_proj",
	"backbone.layers.32.mixer.in_proj",
	"backbone.layers.32.mixer.out_proj",
	"backbone.layers.33.mixer.q_proj",
	"backbone.layers.33.mixer.k_proj",
	"backbone.layers.33.mixer.v_proj",
	"backbone.layers.33.mixer.o_proj",
	"backbone.layers.41.mixer.in_proj",
	"backbone.layers.41.mixer.out_proj",
	"backbone.layers.42.mixer.q_proj",
	"backbone.layers.42.mixer.k_proj",
	"backbone.layers.42.mixer.v_proj",
	"backbone.layers.42.mixer.o_proj",
	"backbone.layers.0.mixer.conv1d",
	"backbone.layers.2.mixer.conv1d",
	"backbone.layers.4.mixer.conv1d",
	"backbone.layers.7.mixer.conv1d",
	"backbone.layers.9.mixer.conv1d",
	"backbone.layers.11.mixer.conv1d",
	"backbone.layers.14.mixer.conv1d",
	"backbone.layers.16.mixer.conv1d",
	"backbone.layers.18.mixer.conv1d",
	"backbone.layers.21.mixer.conv1d",
	"backbone.layers.23.mixer.conv1d",
	"backbone.layers.25.mixer.conv1d",
	"backbone.layers.28.mixer.conv1d",
	"backbone.layers.30.mixer.conv1d",
	"backbone.layers.32.mixer.conv1d",
	"backbone.layers.35.mixer.conv1d",
	"backbone.layers.37.mixer.conv1d",
	"backbone.layers.39.mixer.conv1d",
	"backbone.layers.41.mixer.conv1d",
	"backbone.layers.44.mixer.conv1d",
	"backbone.layers.46.mixer.conv1d",
	"backbone.layers.48.mixer.conv1d",
	"backbone.layers.50.mixer.conv1d"
	]
	}
	}