KevinDu1

Results 3 issues of KevinDu1

I evaluate my custom dataset using `python ./pytorch/train.py evaluate --config_path=./configs/car.fhd.config --model_dir=/path/to/model_dir --measure_time=True --batch_size=1 ` After running, I visualize the results but there are little cars are detected, the 3D bboxes...

感谢大佬精彩的分享,我最近使用Argoverse数据集10步的历史轨迹(1s)去预测30步(3s)的未来轨迹,在训练过程中,前几个epoch结果良好,到了五六个epoch时发生了梯度爆炸,损失函数从原来的0.01升到了10,20,请问,您知道这是怎么回事吗