mqdigihub
commented
2 months ago 可以尝试调小学习率，比如降低到0.001。我训练自己的数据集，也遇到过这种情况，学习率降低到0.001就可以了。 @QMLmulan
Open QMLmulan opened 2 months ago
mqdigihub
commented
2 months ago 可以尝试调小学习率，比如降低到0.001。我训练自己的数据集，也遇到过这种情况，学习率降低到0.001就可以了。 @QMLmulan
QMLmulan
commented
2 months ago 好的，我尝试一下，谢谢~~
------------------ 原始邮件 ------------------ 发件人: "dawn-ech/YOLC" @.>; 发送时间: 2024年8月31日(星期六) 中午11:07 @.>; @.**@.>; 主题: Re: [dawn-ech/YOLC] 训练精度问题 (Issue #9)
可以尝试调小学习率，比如降低到0.001。我训练自己的数据集，也遇到过这种情况，学习率降低到0.001就可以了。 @QMLmulan
— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>
作者您好，在模型训练自己我数据集的时候，在第一个epoch就出现了 2024-08-31 09:14:18,846 - mmdet - INFO - Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=1000 ] = 0.000 Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=1000 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=1000 ] = 0.000 Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=1000 ] = -1.000 Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=1000 ] = -1.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=300 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=1000 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=1000 ] = 0.000 Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=1000 ] = -1.000 Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=1000 ] = -1.000
2024-08-31 09:14:18,849 - mmdet - INFO - +-------------------+-------+ | category | AP | +-------------------+-------+ | Offshore-platform | 0.000 | +-------------------+-------+ 2024-08-31 09:14:19,049 - mmdet - INFO - Exp name: yolc.py 2024-08-31 09:14:19,050 - mmdet - INFO - Epoch(val) [1][624] bbox_mAP: 0.0000, bbox_mAP_50: 0.0000, bbox_mAP_75: 0.0000, bbox_mAP_s: 0.0000, bbox_mAP_m: -1.0000, bbox_mAP_l: -1.0000, bbox_mAP_copypaste: 0.000 0.000 0.000 0.000 -1.000 -1.000 C:\Users\nsd.conda\envs\yolo\lib\site-packages\mmcv\runner\hooks\logger\text.py:112: DeprecationWarning: an integer is required (got type float). Implicit conversion to integers using int is deprecated, and may be removed in a future version of Python. mem_mb = torch.tensor([mem / (1024 * 1024)], 2024-08-31 09:15:00,117 - mmdet - INFO - Epoch [2][50/2130] lr: 2.500e-03, eta: 2:12:38, time: 0.821, data_time: 0.231, memory: 4928, loss_center_heatmap: nan, loss_xywh_coarse: nan, loss_xywh_coarse_l1: nan, loss_xywh_refine: nan, loss_xywh_refine_l1: nan, loss: nan, grad_norm: nan 第二个epoch就是nan 2024-08-31 09:21:10,932 - mmdet - INFO - Epoch [2][650/2130] lr: 2.500e-03, eta: 2:11:30, time: 0.625, data_time: 0.002, memory: 4928, loss_center_heatmap: nan, loss_xywh_coarse: nan, loss_xywh_coarse_l1: nan, loss_xywh_refine: nan, loss_xywh_refine_l1: nan, loss: nan, grad_norm: nan 后面就是一直都是nan，请问这个需要如何解决？