ap值一直是0，当我用dota1.0格式的数据集训练 #34
Description
2022-12-20 18:08:48 | INFO | yolox.core.trainer:244 - epoch: 10/10，iter: 1190/1217,mem: 13929Nb，iter_time: 0.34
4s，data_time: 0.001s， loss_obj: 1.7, loss_cls: 0.9， loss_reg: 1.9，loss_reg_extra: 1.4，total_loss: 5.9，ir: 6.250e-5,size:1024,ETA: 0:00: 16
2022-12-20 18∶08:52 |INFO |yolox.core.trainer:244 - epoch: 10/10，iter: 1200/1217,mem: 13929Nb，iter_time: 0.35
7s，data_time: 0.001s，loss_obj: 1.3, loss_cls: 0.7，loss_reg: 1.7, loss_reg_extra: 1.3，total_loss: 5.0，1r: 6.250e-5, size: 928,ETA: 0:00:10
2022-12-20 18:08:55 |INFO |yolox.core.trainer:244 - epoch: 10/10，iter: 1210/1217，mem: 13929Nb，iter_time: 0.31
1s， data_time: 0.03s，loss_obj: 2.3， loss_cls: 1.1，loss_reg: 1.8， loss_reg_extra: 1.0，total_loss: 6.2，1r: 6.250e-&5,size: 864,ETA: 0:00:04
2022-12-20 18:08:57 |INFO |yolox.core.trainer:320 - Save weights to YOLox_outputs/yoloxs_20221220
2022-12-20 18:08:57 i INFO |yolox.core.trainer:190 - Training of experiment is done and the best AP is 0.00
/root/data
博主您好，我按照您知乎帖子上的方法复现yolox旋转检测，训练80个epoch和10个epoh ap值一直是0，想请教一下您我这个问题可能出现在哪里？