a bug when add data to device

[FlyingDita]

past version：
def to_cuda(self, batch):
for k in batch:
if not isinstance(k, torch.Tensor):
continue
batch[k] = batch[k].cuda()
return batch

it should be:
def to_cuda(self, batch):
for k in batch:
if not isinstance(batch[k], torch.Tensor):
continue
batch[k] = batch[k].cuda()

[zillur-av]

could you please point out the code lines from the repo?

[FlyingDita]

LVLane/lanedet/engine/runner.py line54