RuntimeError: Mismatch in shape

REDHWAN_AL-GABRI · December 19, 2022, 12:43pm

python3 train_hopenet.py --dataset AFLW2000 --data_dir datasets/AFLW2000 --filename_list datasets/AFLW2000/files.txt --output_string er

I got this error:

/home/redhwan/.local/lib/python3.8/site-packages/torch/optim/adam.py:90: UserWarning: optimizer contains a parameter group with duplicate parameters; in future, this will cause an error; see github.com/pytorch/pytorch/issues/40967 for more information
  super(Adam, self).__init__(params, defaults)
Ready to train network.
Traceback (most recent call last):
  File "train_hopenet.py", line 193, in <module>
    torch.autograd.backward(loss_seq, grad_seq)
  File "/home/redhwan/.local/lib/python3.8/site-packages/torch/autograd/__init__.py", line 166, in backward
    grad_tensors_ = _make_grads(tensors, grad_tensors_, is_grads_batched=False)
  File "/home/redhwan/.local/lib/python3.8/site-packages/torch/autograd/__init__.py", line 50, in _make_grads
    raise RuntimeError("Mismatch in shape: grad_output["
RuntimeError: Mismatch in shape: grad_output[0] has a shape of torch.Size([1]) and output[0] has a shape of torch.Size([]).

How can I solve the issue?

torch.__version__ = 1.12.0+cu102

Thanks a lot in advance!

eqy · December 26, 2022, 1:09am

You might want to check if changing this line:

github.com

natanielruiz/deep-head-pose/blob/f7bbb9981c2953c2eca67748d6492a64c8243946/code/train_hopenet.py#L187


      
              loss_reg_yaw = reg_criterion(yaw_predicted, label_yaw_cont)
              loss_reg_pitch = reg_criterion(pitch_predicted, label_pitch_cont)
              loss_reg_roll = reg_criterion(roll_predicted, label_roll_cont)
          
          
    # Total loss
              loss_yaw += alpha * loss_reg_yaw
              loss_pitch += alpha * loss_reg_pitch
              loss_roll += alpha * loss_reg_roll
          
          
    loss_seq = [loss_yaw, loss_pitch, loss_roll]
              grad_seq = [torch.ones(1).cuda(gpu) for _ in range(len(loss_seq))]
              optimizer.zero_grad()
              torch.autograd.backward(loss_seq, grad_seq)
              optimizer.step()
          
          
    if (i+1) % 100 == 0:
                  print ('Epoch [%d/%d], Iter [%d/%d] Losses: Yaw %.4f, Pitch %.4f, Roll %.4f'
                         %(epoch+1, num_epochs, i+1, len(pose_dataset)//batch_size, loss_yaw.data[0], loss_pitch.data[0], loss_roll.data[0]))
          
          
# Save models at numbered epochs.
          if epoch % 1 == 0 and epoch < num_epochs:

to grad_seq = [torch.ones([1]).cuda(gpu) for _ in range(len(loss_seq))] (note the added brackets) resolves the issue.

REDHWAN_AL-GABRI · December 26, 2022, 1:32am

First of all, thank you so much for your help.

Your suggestion didn’t solve the issue.

It is working well by:

grad_seq = [torch.tensor(1, dtype=torch.float) .cuda(gpu) for _ in range(len(loss_seq))]