search

ultralytics / ultralytics

NEW - YOLOv8 🚀 in PyTorch > ONNX > OpenVINO > CoreML > TFLite
https://docs.ultralytics.com
GNU Affero General Public License v3.0
24.46k stars 4.86k forks source link

Train the selected model structure for your dataset #12792

Open smallMantou opened 2 weeks ago

smallMantou commented 2 weeks ago

Search before asking

Question

Hello, I have some questions about training my own dataset. I want to start training my dataset from scratch. How do I know if my model is based on n, s, m, or l? How can I make changes to train S? Does the following image indicate that I am based on n

n

Additional

No response

deepukr007 commented 2 weeks ago

I think AMP check always happens on yolov8n, I see the same thing everytime.. You can ignore it. If your YAML file has name such as yolov8s_{....} .yaml your model will be based on yolov8s.. You can also cross check this by looking at the total model parameters .

smallMantou commented 1 week ago

How can I change it to be based on YOLOv8S?

glenn-jocher commented 1 week ago

Hello! To switch your model base to YOLOv8s, you'll need to use a YOLOv8s model configuration file. You can specify this in your training command by replacing the model argument with the path to a YOLOv8s YAML file. For example:

yolo detect train data=your_dataset.yaml model=yolov8s.yaml epochs=100 imgsz=640

Make sure you have the yolov8s.yaml file in your directory, or you can download it from the Ultralytics repository if it's not already included. This will set up your training to use the YOLOv8s architecture. Happy training! 🚀