Instructions to use nn-tech/MetalGPT-1-FP8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nn-tech/MetalGPT-1-FP8 with Transformers:

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

pipe = pipeline("text-generation", model="nn-tech/MetalGPT-1-FP8")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nn-tech/MetalGPT-1-FP8")
model = AutoModelForCausalLM.from_pretrained("nn-tech/MetalGPT-1-FP8")
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 nn-tech/MetalGPT-1-FP8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nn-tech/MetalGPT-1-FP8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nn-tech/MetalGPT-1-FP8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nn-tech/MetalGPT-1-FP8

SGLang

How to use nn-tech/MetalGPT-1-FP8 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 "nn-tech/MetalGPT-1-FP8" \
    --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": "nn-tech/MetalGPT-1-FP8",
		"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 "nn-tech/MetalGPT-1-FP8" \
        --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": "nn-tech/MetalGPT-1-FP8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nn-tech/MetalGPT-1-FP8 with Docker Model Runner:
```
docker model run hf.co/nn-tech/MetalGPT-1-FP8
```

Description

MetalGPT-1 is a model built upon the Qwen/Qwen3-32B and incorporates both continual pre-training and supervised fine-tuning on domain-specific data from the mining and metallurgy industry.

Quantization

For convenience and improved performance, we also provide this FP8 checkpoint of the nn-tech/MetalGPT-1 model. Using FP8 precision enables faster inference and lower memory usage, while preserving model quality and numerical stability.

VLLM usage

vllm serve nn-tech/MetalGPT-1-FP8 --reasoning-parser qwen3


from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:8000/v1",
    api_key="dummy"  
)

response = client.chat.completions.create(
    model="nn-tech/MetalGPT-1-FP8",
    messages=[
        {"role": "system", "content": "Ты специалист в области металлургии."},
        {"role": "user", "content": "Назови плюсы и минусы хлоридной и сульфатной технологии производства никеля."}
    ],
    temperature=0.7,
    max_tokens=1024
)

print(response.choices[0].message.content)

Downloads last month: 87

Safetensors

Model size

33B params

Tensor type

BF16

F8_E4M3

Model tree for nn-tech/MetalGPT-1-FP8

Base model

Qwen/Qwen3-32B

Finetuned

t-tech/T-pro-it-2.0

Finetuned

nn-tech/MetalGPT-1

Quantized

(3)

this model