Daniel Ji

yes, I follow the author's envs, and I complete it successfully.

learning rate is too large, plz try small lr

可能永远不会再由 @QianMo 扩充了

I guess the issue coming from the CUDA version. What's your version of CUDA and pytorch on V100?

@StevenGrove Thank you again.

And what's a difference between [RSIS](https://github.com/imatge-upc/rvos/blob/f8f8b1039bf2808b7d22fdb050b2f06ac6a5b172/src/modules/model.py#L70) and [RSISMask](https://github.com/imatge-upc/rvos/blob/f8f8b1039bf2808b7d22fdb050b2f06ac6a5b172/src/modules/model.py#L170)? Similarly, what's the difference between [ConvLSTMCell](https://github.com/imatge-upc/rvos/blob/f8f8b1039bf2808b7d22fdb050b2f06ac6a5b172/src/modules/clstm.py#L7) and [ConvLSTMCellMask](https://github.com/imatge-upc/rvos/blob/f8f8b1039bf2808b7d22fdb050b2f06ac6a5b172/src/modules/clstm.py#L71)? thank your kindness reply! @

谢谢作者亲答。 我还有一个疑问就是，你说的GC在前面和后面的区别在于？

我在tf-1.6和python3下跑通了代码，修改了很多地方，已解决问题，谢谢了

@mahdiyehbakhshi822 Hi, res2net weight file will be automatically downloaded by this [line](https://github.com/DengPingFan/PraNet/blob/ba45d69dccbf73d131ecad5f429fec82ff56d790/jittor/lib/Res2Net_v1b.py#L8)

Hi, @Moqixis @zzzzzzpc I merge the response in another issue. I hope the answer helps you guys a lot. Best, Ge-Peng.