Gavin666Github

@huanghoujing @ShiinaMitsuki 谢谢指导！应该是内存不够大的前提下跑不起来，out of memory,其实我这边本机(单gpu)内存只有4G，单跑Global Loss是没问题的(batch size调的比较小)，但是同样的batch size跑 mutual learning 就不行了，直接换到服务器多个GPU跑了ok的，每个GPU分配的Memory =8G ,完美解决。

上个图看看，两个GPU跑的效果  

> > I have updated my create_dataset files for more details about the annotation file hope you will get it. > > Really really appreciate that!!! Actually I have read...

> I have updated my create_dataset files for more details about the annotation file hope you will get it. Hi,G,could u please share me the fall-detection dataset ?the links dead.my...

> 你加我QQ吧，我发你406174050 刚加，感谢

> > > 你加我QQ吧，我发你406174050 > > > > > > 刚加，感谢 > > 能否加我qq也发我一份啊，感谢感谢！951357158 我加你Q

same problem,anybody who can provide baidu cloud disk link to download the UCF101 rgb and flow dataset ,many thanks!

hi,you can do like this: python3 train.py -g 0 --load_model result/model1000.npz