Hao Wei

你好，已收到你的邮件。                                                        --魏浩

@Pan-zhaoyu Thank you. I have met the overfitting issue in train the network. And could you tell me how to fix the issue? I will very appreciate it if you...

I have uploaded these trained models in Baidu Driver. 链接：https://pan.baidu.com/s/1ssqs67E7moo8a-ix61daLg 提取码：t0rn

you should rename the directory "relaynet_pytorch" to "networks" !

你好，已收到你的邮件。                                                        --魏浩

@manyu90 I think your view is correct and thank you for your supplements.

you can find the classification pre-trained model at [here](https://drive.google.com/drive/folders/1eC9xBexK4aKoNzPRiLU2sYu_UyYm3TRU), but only including the model of setting in static_24 and static_64, without the model of dynamic setting (input channel=192). I need...

Yes, I can download these pre-training models completely and also can load these weights correctly into the model. Maybe your network is unstable.

you can find the classification pre-trained model at [here](https://drive.google.com/drive/folders/1eC9xBexK4aKoNzPRiLU2sYu_UyYm3TRU), but only including the model of setting in static_24 and static_64, without the model of dynamic setting (input channel=192). I need...

我刚开始也遇到这个问题，后来将sat的版本从0.4.9降低到0.3.6，然后将transoformers的版本定在4.27.1，就没有错误了。 原因应该是checkpoint是旧版本的，不兼容新版本的sat。 希望对你有用~