AttributeError: ‘str‘ object has no attribute ‘param_group‘

DeamNet 代码调试

 Epoch=2000，运行到999次报错AttributeError: 'str' object has no attribute 'param_group'

Traceback (most recent call last):
  File "D:/Papers to read/2022.07/DeamNet-main/DeamNet-main/train.py", line 189, in <module>
    if (epoch + 1) % (opt.nEpochs / 2) == 0:
AttributeError: 'str' object has no attribute 'param_group'

Process finished with exit code 1

错误代码段： 

 if (epoch + 1) % (opt.nEpochs / 2) == 0:
            for param_group in optimizer.param_groups:
                param_group['lr'] /= 10.0
            print('Learning rate decay: lr={}'.param_group['lr'])

optimizer.param_groups： 是长度为2的list，其中的元素是2个字典；optimizer.param_groups[0]： 长度为6的字典，包括[‘amsgrad’, ‘params’, ‘lr’, ‘betas’, ‘weight_decay’, ‘eps’]这6个参数；
optimizer.param_groups[1]： 好像是表示优化器状态的一个字典.

没找到好的解决办法，所以结合看到的print内容，改代码为：

      if (epoch + 1) % (opt.nEpochs / 2) == 0:
            for param_group in optimizer.param_groups:
                param_group['lr'] /= 10.0
            # print('Learning rate decay: lr={}'.param_group['lr'])
            print('Learning rate decay: lr={}')

由于源代码设置epoch=2000, 运行比较慢，我只是想确定代码没有错误，所以改为epoch=10，简单测试一下。现在不报错了 

暂时不确定这行代码不打印后半部分的影响，只是想测试代码，所以算是跑通了，准备开始学习代码主要内容啦。

Learning rate decay: lr={}

===> Epoch[10](1/20): Loss: 0.0002 || Timer: 0.0370 sec.
===> Epoch[10](2/20): Loss: 0.0006 || Timer: 0.0328 sec.
.................
===> Epoch[10](19/20): Loss: 0.0004 || Timer: 0.0358 sec.
===> Epoch[10](20/20): Loss: 0.0003 || Timer: 0.0347 sec.
===> Epoch 10 Complete: Avg. Loss: 0.0004

===> Avg. PSNR: 30.5941 dB

Process finished with exit code 0