Instructions to use nvidia/Llama-3.1-70B-Instruct-FP8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nvidia/Llama-3.1-70B-Instruct-FP8 with Transformers:

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

pipe = pipeline("text-generation", model="nvidia/Llama-3.1-70B-Instruct-FP8")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nvidia/Llama-3.1-70B-Instruct-FP8")
model = AutoModelForCausalLM.from_pretrained("nvidia/Llama-3.1-70B-Instruct-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 Settings

vLLM

How to use nvidia/Llama-3.1-70B-Instruct-FP8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nvidia/Llama-3.1-70B-Instruct-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": "nvidia/Llama-3.1-70B-Instruct-FP8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nvidia/Llama-3.1-70B-Instruct-FP8

SGLang

How to use nvidia/Llama-3.1-70B-Instruct-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 "nvidia/Llama-3.1-70B-Instruct-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": "nvidia/Llama-3.1-70B-Instruct-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 "nvidia/Llama-3.1-70B-Instruct-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": "nvidia/Llama-3.1-70B-Instruct-FP8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nvidia/Llama-3.1-70B-Instruct-FP8 with Docker Model Runner:
```
docker model run hf.co/nvidia/Llama-3.1-70B-Instruct-FP8
```

License incompatibility

by qiuqiu666 - opened Jun 25, 2025

Discussion

qiuqiu666

Jun 25, 2025

Hi , I’d like to report a potential license conflict in nvidia/Llama-3.1-70B-Instruct-FP8. This model appears to be fine-tuned from
meta-llama/Llama-3.1-70B-Instruct, which is governed by the LLaMA 3.1 Community License.

However, the fine-tuned model is currently being distributed under the NVIDIA Open Model License, which may not comply with the restrictions and redistribution clauses in the original Meta license.

⚠️ Key violations of META LLAMA 3.1 COMMUNITY LICENSE AGREEMENT:

Clause 1.b – Redistribution and Use:
  • All derivatives must retain the original license
  • You may not sublicense under a different license
  • You must include the LLaMA license and its usage restrictions

Clause 1.c – Acceptable Use:
  • Downstream usage must follow Meta’s Acceptable Use Policy
  • Redistribution must not relax these obligations

Clause 2 – No Additional Grant:
  • No extra permissions or terms beyond what Meta grants are allowed

Clause 3 – Commercial Restrictions:
  • Use at commercial scale (e.g. >700M MAU) requires additional license

Your NVIDIA Open Model License includes:

  • Broad commercial rights
  • Right to sublicense and add terms for derivative models
  • No requirement to retain the LLaMA license text
  • NVIDIA branding and redistribution conditions (e.g., “Built on NVIDIA Cosmos”)

To align with the upstream LLaMA 3.1 license:

1. Restore the LLaMA 3.1 Community License as the primary license in the model card

2. Include this NOTICE in a file or the model card:
 > "This model is based on LLaMA 3.1 and is licensed under the LLaMA 3.1 Community License, Copyright © Meta Platforms, Inc."

3. Remove or clarify NVIDIA-specific terms that conflict with Meta’s redistribution limits

4. Make clear that any downstream use must comply with Meta’s Acceptable Use Policy

Hope this helps! 😊 Let me know if you have any questions or need more info.

Thanks for your attention!

qiuqiu666

Jul 10, 2025

Would love to hear your view on this!

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