Loss does not converge during training, which results in a poor evaluation results(map:41.92%)

hunglc007 / tensorflow-yolov4-tflite

YOLOv4, YOLOv4-tiny, YOLOv3, YOLOv3-tiny Implemented in Tensorflow 2.0, Android. Convert YOLO v4 .weights tensorflow, tensorrt and tflite

https://github.com/hunglc007/tensorflow-yolov4-tflite

MIT License

2.23k stars 1.24k forks source link

Loss does not converge during training, which results in a poor evaluation results(map:41.92%) #394

Open Sebastian514 opened 2 years ago

Sebastian514 commented 2 years ago

I used the default hyper-parameters and trained the model on a single Nvidia Titan Xp. And the training logs show that the loss
didn't converge. It apparently leaded to a bad performance which evaluated on coco.

What should I do?