can anyone guide to do quantization for custom trained yolov8 about ultralytics

[github-actions]

👋 Hello @218w1d7706, thank you for your interest in Ultralytics YOLOv8 🚀! We recommend a visit to the Docs for new users where you can find many Python and CLI usage examples and where many of the most common questions may already be answered.

If this is a 🐛 Bug Report, please provide a minimum reproducible example to help us debug it.

If this is a custom training ❓ Question, please provide as much information as possible, including dataset image examples and training logs, and verify you are following our Tips for Best Training Results.

Join the vibrant Ultralytics Discord 🎧 community for real-time conversations and collaborations. This platform offers a perfect space to inquire, showcase your work, and connect with fellow Ultralytics users.

Install

Pip install the ultralytics package including all requirements in a Python>=3.8 environment with PyTorch>=1.8.

pip install ultralytics

Environments

YOLOv8 may be run in any of the following up-to-date verified environments (with all dependencies including CUDA/CUDNN, Python and PyTorch preinstalled):

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

If this badge is green, all Ultralytics CI tests are currently passing. CI tests verify correct operation of all YOLOv8 Modes and Tasks on macOS, Windows, and Ubuntu every 24 hours and on every commit.

[glenn-jocher]

Hello! For quantizing a custom-trained YOLOv8 model, you can use the export functionality with INT8 quantization. Here's a simple example on how to export your model to TensorRT format with INT8 precision:

yolo export model=path/to/your/custom_model.pt format=engine int8=True

Make sure to perform this on the same device you plan to deploy the model, as INT8 calibration is device-specific. For more detailed guidance, you can refer to the TensorRT integration documentation provided by Ultralytics.

If you encounter any specific issues during this process, feel free to share them here for more targeted assistance! 🚀

[the0807]

I'm trying to reduce the loss, but try it
https://github.com/the0807/YOLOv8-ONNX-TensorRT

[glenn-jocher]

Hello! To reduce the loss during training, ensure your dataset is well-prepped and consider tweaking hyperparameters like learning rate or batch size. Also, using a pre-trained model can provide a good starting point. For specific adjustments in loss, reviewing the training logs to understand where the model might be underperforming can be helpful. If you're looking into using TensorRT for optimization, ensure your model is properly calibrated, especially when using INT8 precision. Good luck! 🚀