Xu Jing zhao

Results 83 comments of Xu Jing zhao

Apparently there is a bug in pytorch , when you open a new dataloader , it seems that the older dataloader will not be released , I have meet this...

I don't think it's crucial whether there exist a relu function after the first conv.

> Can you upload the dataset on Google cloud or anything, because Baidu drive is unaccessible here. HH , I think he can not access Google cloud also.

> > 链接:https://pan.baidu.com/s/1dP1mXuU-rA9dPvFe8YS8jQ > > 提取码:k6rc > > 这个解压有问题啊!! 你能解压出来? 那个是tensorflow版本的吧,老哥pytorch版本的有吗

我改了dataset,只有在getittem的时候才读取Img,而不是一开始就读取img

> @heartInsert 你每次get_item再去读取图片,会从某个角度影响性能,决定于磁盘io。如果你多个worker还行,如果不是的话,就需要预加载,就是让后台去提前加载这部分图像到内存,然后到时内存直接读取,而不需要等待了。 00,对的,这样确实比较慢,直接读内存那种肯定快好多

应该是Loss里面作了处以0的操作,你clamp一下