Error in DualTaskLoss while running the evaluation

[shabnambanu1983]

Hi Coders,

    I am trying to run the evaluation on the cityscapes databases as discussed in the README  . I am using 2 GPUs, did all the setups and downloaded the cityscapes datasets as required by the setup.

 After crossing the basic steps, now its failing in the DuakTaskLoss. I am pasting the error for your reference. Please help me with the issue. Thanks in advance.

=====================================error Log=============================================

 shabnamenv) root@shabnam2-gpu:/temp/WORKSPACE/GSCNN# python train.py --evaluate --snapshot checkpoints/best_cityscapes_checkpoint.pth

/root/SETUPS/anaconda3/envs/shabnamenv/lib/python3.8/site-packages/setuptools/distutils_patch.py:25: UserWarning: Distutils was imported before Setuptools. This usage is discouraged and may exhibit undesirable behaviors or errors. Please use Setuptools' objects directly or at least import Setuptools first. warnings.warn( 08-14 16:05:36.060 train fine cities: ['train/aachen', 'train/bochum', 'train/bremen', 'train/cologne', 'train/darmstadt', 'train/dusseldorf', 'train/erfurt', 'train/hamburg', 'train/hanover', 'train/jena', 'train/krefeld', 'train/monchengladbach', 'train/strasbourg', 'train/stuttgart', 'train/tubingen', 'train/ulm', 'train/weimar', 'train/zurich'] 08-14 16:05:36.071 Cityscapes-train: 2975 images 08-14 16:05:36.071 val fine cities: ['val/frankfurt', 'val/munster', 'val/lindau'] 08-14 16:05:36.073 Cityscapes-val: 500 images 08-14 16:05:36.073 Using Per Image based weighted loss /root/SETUPS/anaconda3/envs/shabnamenv/lib/python3.8/site-packages/torch/nn/modules/loss.py:217: UserWarning: NLLLoss2d has been deprecated. Please use NLLLoss instead as a drop-in replacement and see https://pytorch.org/docs/master/nn.html#torch.nn.NLLLoss for more details. warnings.warn("NLLLoss2d has been deprecated. " /root/SETUPS/anaconda3/envs/shabnamenv/lib/python3.8/site-packages/torch/nn/reduction.py:44: UserWarning: size_average and reduce args will be deprecated, please use reduction='mean' instead. warnings.warn(warning.format(ret)) 08-14 16:05:36.074 Using Cross Entropy Loss /root/SETUPS/anaconda3/envs/shabnamenv/lib/python3.8/site-packages/encoding/nn/syncbn.py:228: EncodingDeprecationWarning: encoding.nn.BatchNorm2d is now deprecated in favor of encoding.nn.SyncBatchNorm. warnings.warn("encoding.nn.{} is now deprecated in favor of encoding.nn.{}." /temp/WORKSPACE/GSCNN/network/mynn.py:29: UserWarning: nn.init.kaiming_normal is now deprecated in favor of nn.init.kaiming_normal. nn.init.kaiming_normal(module.weight) 08-14 16:05:37.111 Model params = 137.3M 08-14 16:05:39.992 Loading weights from model checkpoints/best_cityscapes_checkpoint.pth 08-14 16:05:40.656 Load Compelete /root/SETUPS/anaconda3/envs/shabnamenv/lib/python3.8/site-packages/torch/nn/functional.py:3118: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details. warnings.warn("Default upsampling behavior when mode={} is changed " /temp/WORKSPACE/GSCNN/loss.py:160: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument. return self.nll_loss(F.log_softmax(inputs), targets) Traceback (most recent call last): File "train.py", line 381, in main() File "train.py", line 140, in main validate(val_loader, net, criterion_val, File "train.py", line 303, in validate loss_dict = criterion((seg_out, edge_out), (mask_cuda, edge_cuda)) File "/root/SETUPS/anaconda3/envs/shabnamenv/lib/python3.8/site-packages/torch/nn/modules/module.py", line 722, in _call_impl result = self.forward(*input, **kwargs) File "/temp/WORKSPACE/GSCNN/loss.py", line 109, in forward losses['dual_loss'] = self.dual_weight * self.dual_task(segin, segmask) File "/root/SETUPS/anaconda3/envs/shabnamenv/lib/python3.8/site-packages/torch/nn/modules/module.py", line 722, in _call_impl result = self.forward(*input, **kwargs) File "/temp/WORKSPACE/GSCNN/my_functionals/DualTaskLoss.py", line 120, in forward g_hat = g_hat.view(N, -1) RuntimeError: view size is not compatible with input tensor's size and stride (at least one dimension spans across two contiguous subspaces). Use .reshape(...) instead.

Thanks, Shabnam

[iariav]

just to as the massage suggests - replace g_hat = g_hat.view(N, -1) with g_hat = g_hat.reshape(N, -1)