Yuhui Ma
Yuhui Ma
In the code, loaded samples seem only to substract mean but not to be normalized, which leads to very large loss. Why not normalize the samples?
您好！我这边用这个代码处理单张图没问题，但python里面写for循环批量处理图片的时候GPU的显存占用会随着处理图片数量的增大而增大，最终会报CUDA out of memory，请问该如何解决？
How to deal with the problem of nan loss during training?
Hello, could you tell me how to run your tensorflow code on my own datasets?