huihuiustc
huihuiustc
What dou mean that you trained the large model from smaller ones.Or this one:"We initialize deeper networks by parameters from shallower ones for faster convergence"in your paper. For instance. We...
谢谢大佬的回复,确实是很厉害的工作和研究。 我们正在尝试先把可变卷积换成正常的卷积,然后训练得到的初始model,然后用这个模型训练网络。 接着冻结部分模型块再开始训练。
Could you share me the leftImg8bit_trainvaltest_foggy.zip ?
好像是图像的尺寸不一样,我们是不是应该补齐图像到同样的尺寸?
i also have the same problem, have you solve it?
It happens when testing
现在还是会nan,我以为是我自己写的模型的有问题,后换成官方的源代码,官方提供的预训练模型,训练1轮以后开始出现nan。
Any update?
I need it too