How to train without cfg

ultralytics / yolov3

YOLOv3 in PyTorch > ONNX > CoreML > TFLite

https://docs.ultralytics.com

GNU Affero General Public License v3.0

10.18k stars 3.44k forks source link

How to train without cfg #1208

Closed litingsjj closed 4 years ago

litingsjj commented 4 years ago

Did you have plan to write example without cfg? just use pytorch create nn

glenn-jocher commented 4 years ago

@litingsjj we are planning this for our next repository. For this repo you must use a cfg.

dsp6414 commented 4 years ago

this a good idea, Modular design idea ,similar to detectron 2, user can freely change backbone and loss (Giou,Diou,Ciou)

litingsjj commented 4 years ago

So except...

glenn-jocher commented 11 months ago

@litingsjj Yes, that's a great point. We're always looking for ways to improve functionality, and we appreciate your input. If you have any more ideas or suggestions, we'd love to hear them!

dsp6414 commented 11 months ago

已收到！

glenn-jocher commented 11 months ago

@dsp6414 glad to hear that! If you have any other questions or suggestions, feel free to reach out. Happy to help!