Llama 3.1 8B Instruct FP8

🚀 NVIDIA Llama 3.1 8B Instruct FP8 Model

The NVIDIA Llama 3.1 8B Instruct FP8 model is a quantized version of Meta's Llama 3.1 8B Instruct model, offering efficient text generation capabilities.

Metadata

🚀 Quick Start

The NVIDIA Llama 3.1 8B Instruct FP8 model is the quantized version of the Meta's Llama 3.1 8B Instruct model, which is an auto - regressive language model using an optimized transformer architecture. For more information, check here. It is quantized with TensorRT Model Optimizer. This model is ready for both commercial and non - commercial use.

✨ Features

• Quantized Model: Reduces disk size and GPU memory requirements by approximately 50% through post - training quantization to FP8.
• Multiple Runtime Support: Compatible with Tensor(RT) - LLM and vLLM for inference.
• Wide Hardware Compatibility: Supports NVIDIA Blackwell, Hopper, and Lovelace microarchitectures.

📦 Installation

Deploy with TensorRT - LLM

To deploy the quantized checkpoint with TensorRT - LLM, follow these steps:

1. Checkpoint convertion:

python examples/llama/convert_checkpoint.py --model_dir Llama-3.1-8B-Instruct-FP8 --output_dir /ckpt --use_fp8

2. Build engines:

trtllm-build --checkpoint_dir /ckpt --output_dir /engine

3. Throughputs evaluation: Refer to the TensorRT - LLM benchmarking documentation for details.

Deploy with vLLM

To deploy the quantized checkpoint with vLLM, follow these instructions:

1. Install vLLM from directions here.
2. When using a Model Optimizer PTQ checkpoint with vLLM, pass the quantization = modelopt flag into the config while initializing the LLM Engine.

💻 Usage Examples

Basic Usage

# Example deployment on H100 with vLLM
from vllm import LLM, SamplingParams

model_id = "nvidia/Llama-3.1-8B-Instruct-FP8"
sampling_params = SamplingParams(temperature=0.8, top_p=0.9)

prompts = [
    "Hello, my name is",
    "The president of the United States is",
    "The capital of France is",
    "The future of AI is",
]

llm = LLM(model=model_id, quantization="modelopt")
outputs = llm.generate(prompts, sampling_params)

# Print the outputs.
for output in outputs:
    prompt = output.prompt
    generated_text = output.outputs[0].text
    print(f"Prompt: {prompt!r}, Generated text: {generated_text!r}")

📚 Documentation

Model Overview

Third - Party Community Consideration

This model is not owned or developed by NVIDIA. It has been developed and built to a third - party’s requirements for this application and use case. See the link to the Non - NVIDIA (Meta - Llama - 3.1 - 8B - Instruct) Model Card.

License/Terms of Use

• nvidia - open - model - license
• llama3.1

Model Architecture

• Architecture Type: Transformers
• Network Architecture: Llama3.1

Input

• Input Type(s): Text
• Input Format(s): String
• Input Parameters: Sequences
• Other Properties Related to Input: Context length up to 128K

Output

• Output Type(s): Text
• Output Format: String
• Output Parameters: Sequences
• Other Properties Related to Output: N/A

Software Integration

• Supported Runtime Engine(s):
  • Tensor(RT) - LLM
  • vLLM
• Supported Hardware Microarchitecture Compatibility:
  • NVIDIA Blackwell
  • NVIDIA Hopper
  • NVIDIA Lovelace
• Preferred Operating System(s):
  • Linux

Model Version(s)

The model is quantized with nvidia - modelopt v0.27.0

Datasets

• Calibration Dataset: cnn_dailymail
• Evaluation Dataset: MMLU

Inference

• Engine: Tensor(RT) - LLM or vLLM
• Test Hardware: H100

Post Training Quantization

This model was obtained by quantizing the weights and activations of Meta - Llama - 3.1 - 8B - Instruct to FP8 data type, ready for inference with TensorRT - LLM and vLLM. Only the weights and activations of the linear operators within transformers blocks are quantized. This optimization reduces the number of bits per parameter from 16 to 8, reducing the disk size and GPU memory requirements by approximately 50%. On H100, a 1.3x speedup was achieved.

Evaluation

We benchmarked with tensorrt - llm v0.13 on 8 H100 GPUs, using a batch size of 1024 for throughputs with in - flight batching enabled. An approximately ~1.3x speedup was achieved with FP8.

Deploy with vLLM

This model can be deployed with an OpenAI Compatible Server via the vLLM backend. See instructions here.

📄 License

• nvidia - open - model - license
• llama3.1