uber-research / LaneGCN

[ECCV2020 Oral] Learning Lane Graph Representations for Motion Forecasting
https://arxiv.org/abs/2007.13732
Other
507 stars 131 forks source link

dilated_nbrs bug #14

Open strongdeer opened 3 years ago

strongdeer commented 3 years ago

Is it that mat = mat * mat should be modified to mat = mat * csr in function dilated_nbrs?

kaikaishen commented 3 years ago

it should not be modified to mat = mat * csr, because of double expansion. I am deploying model to C++ Project.Welcome to exchange.

zyandtom commented 3 years ago

@kaikaishen hi have you convert this model to c++ sucessfully? actually I'm doing similar thing, I try to export the pt file in torch.

kaikaishen commented 3 years ago

@kaikaishen hi have you convert this model to c++ sucessfully? actually I'm doing similar thing, I try to export the pt file in torch.

sucessfully,we can talk about it.my email is 15195875959@163.com