Nihal Gupta

icon company Delhi Technological University icon location Delhi icon location I am a research scholar, working in the field of optical flow and rearrangement tasks.

Results 5 comments of Nihal Gupta

PyTorch1.10+cu113 CUDA Graph for torch_scatter raises RuntimeError w/o argument `out`

which version of torch scatter, are you using with pytorch 1.10

problem with depth image

yes

ISSUE with RPKNET: not able to train

> Hello, thanks for reporting. > > From your log, it looks like the error happens in the middle of the training, after 400 batches, is that correct? If so,...

ISSUE with RPKNET: not able to train

> Hello. I was able to complete one epoch of training using two GPUs without any errors, so I am not sure what is causing your problem. I have two...

ISSUE with RPKNET: not able to train

i did set cuda 12.1, then tried still same issue as u asked printed the out_ch and slice_idx weight shape: torch.Size([256, 1, 3, 3]) x shape: torch.Size([10, 256, 44, 60])...