tensorboy
tensorboy
with torch.no_grad() : output= model(image) may help.
Or you should buy a GPU with large memory. :)
It's strange, it is super quick on 1080ti...
It is weird that operation takes so much time......
It's a little bit harder for me to figure it out, I'm still thinking about it.
Hi, @Naruto-Zhao Hope you are doing well. I'm back to finding the problem. :)
有更新吗这个需求
Hi, @Vaufreyd The mask generated by using original genCOCOMask.m. :)
That may helps: https://discuss.pytorch.org/t/solved-rtx-2080-ti-cudnn-error-occurs-when-loading-rnn-model-to-gpu/33880/12
Hi @hyli2020 This is mainly to respect the original implementation, I realized that there are lots of redundant in the code long time ago, but no time to optimize further,...