VisualDL log change curve

[Winnie202]

Thanks for your code contribution, the following is my relevant training environment :

platform: Linux-5.15.0-43-generic-x86_64-with-debian-bullseye-sid Python: 3.6.13 |Anaconda, Inc.| (default, Jun 4 2021, 14:25:59) [GCC 7.5.0] Paddle compiled with cuda: True NVCC: Build cuda_11.3.r11.3/compiler.29920130_0 cudnn: 8.2 GPUs used: 1 CUDA_VISIBLE_DEVICES: 0 GPU: ['GPU 0: NVIDIA GeForce', 'GPU 1: NVIDIA GeForce'] GCC: gcc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0 PaddleSeg: 2.6.0 PaddlePaddle: 2.2.2 OpenCV: 4.2.0

I use ocrnet to train my task, but the trend doesn't feel normal, please help me

[shiyutang]

It seems that your training process is broken. Can you train from start and provide the whole vdl which includes the validation metric? I can not see anything wrong in the training config. Some of the parameters is relevant to your data.

[Winnie202]

看来你的训练过程被打破了。您可以从头开始训练并提供包含验证指标的整个 vdl 吗？我在训练配置中看不到任何错误。一些参数与您的数据相关。

Because cuda memory is not enough, it is interrupted. Is it normal that the value of train/lr does not change?

[shiyutang]

No, This is not normal. If your cuda memory is not enough, you can try to use smaller batchsize.

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed in 7 days if no further activity occurs. Thank you for your contributions.