Error in evaluation script

[msd-mohit]

Hi,

Thanks for uploading the code, I tried to run the evaluation script by following the readme:
python exp.py --net resunet3.16_penone.dirs.avg.seq+9+1+unet+5+2+16.single+mlpdir+mean+3+64+16 --cmd eval --iter last --eval-dsets tat-subseq

and got the following error:
[2021-11-18/13:38/INFO/exp] create target images
Traceback (most recent call last):
File "exp.py", line 945, in
worker.do(args, worker_objects)
File "../co/mytorch.py", line 442, in do
self.do_cmd(args, worker_objects)
File "../co/mytorch.py", line 430, in do_cmd
worker_objects, iters=args.iter, net_root=args.eval_net_root
File "../co/mytorch.py", line 621, in eval_iters
self.eval(iter, net, eval_sets)
File "../co/mytorch.py", line 630, in eval
self.eval_set(iter, net, eval_set_idx, eval_set, epoch=epoch)
File "exp.py", line 540, in eval_set
self.my_eval_set(iter, net, eval_set_idx, eval_set, epoch)
File "exp.py", line 666, in my_eval_set
output, self.eval_batch_size, height, width, **self.data
File "/user/StableViewSynthesis/experiments/modules.py", line 759, in ref_net_forward
return self.ref_net(ptx, bs, height, width, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/torch/nn/modules/module.py", line 889, in _call_impl
result = self.forward(*input, **kwargs)
File "/user/StableViewSynthesis/experiments/modules.py", line 477, in forward
x = torch_scatter.segment_csr(ptx, point_key, reduce="mean")
File "/usr/local/lib/python3.6/dist-packages/torch_scatter/segment_csr.py", line 103, in segment_csr
return segment_mean_csr(src, indptr, out)
File "/usr/local/lib/python3.6/dist-packages/torch_scatter/segment_csr.py", line 18, in segment_mean_csr
return torch.ops.torch_scatter.segment_mean_csr(src, indptr, out)
RuntimeError: Not compiled with CUDA support

Could you help in this regard?

[griegler]

return torch.ops.torch_scatter.segment_mean_csr(src, indptr, out)
RuntimeError: Not compiled with CUDA support

Looks like you try to use torch with cuda, but the version of PyTorch you are using has no cuda support.