Changing backbone to ResNet-18 about erasing-attention-consistency HOT 2 OPEN

请问为什么我直接使用您提供的代码，模型结构方面没有做任何修改，并直接使用了链接中提供的ResNet50的预训练模型，在clean标签上训练得到的测试集的最好结果仅为82%左右?

from erasing-attention-consistency.

学习率需要从1e-4调整到2e-4。
这个不太可能，估计是你代码某个地方出了问题，或者标签没有读对，82%的准确率过于低了。
你可以参考一下我对其他人问题的回复，如下：
“I have re-run my code without making any changes. Using the pre-trained ResNet-50 model, it easily achieved an accuracy of 90.35% and the best accuracy recorded was 90.51%, without any label noise. Could you please either re-run my code or provide me with your log file to assist me in identifying the reason behind this?”

from erasing-attention-consistency.