runtime error in haca3-test version 1.0.7 related to dimensions

[ibrwr]

Running haca3-test with tag 1.0.4 runs without any problems, but when repeating the same command with 1.0.7 the script gives a runtime error related to dimensions, see below:

haca3-test --in-path $(pwd)/sub-75_ses-tp1_acq-ge_run-1_T1w_2_hacamni.nii.gz --target-theta 0.306601 9.588893 --harmonization-model /somepath/apptainer/haca3/data/harmonization_public.pt --fusion-model /somepath/apptainer/haca3/data/fusion.pt --out-path siemens2hac3 --intermediate-out-dir ./tmp/
========== BEGIN HACA3 HARMONIZATION ==========
Traceback (most recent call last):
File "/somepath/haca3/haca3_venv/bin/haca3-test", line 8, in
sys.exit(main())
File "/somepath/haca3/haca3_venv/lib64/python3.9/site-packages/haca3/test.py", line 120, in main
haca3 = HACA3(beta_dim=args.beta_dim,
File "/somepath/haca3/haca3_venv/lib64/python3.9/site-packages/haca3/modules/model.py", line 53, in init
self.decoder.load_state_dict(self.checkpoint['decoder'])
File "/somepath/haca3/haca3_venv/lib64/python3.9/site-packages/torch/nn/modules/module.py", line 2152, in load_state_dict
raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format(
RuntimeError: Error(s) in loading state_dict for UNet:
size mismatch for in_conv.weight: copying a param with shape torch.Size([8, 3, 3, 3]) from checkpoint, the shape in current model is torch.Size([16, 3, 3, 3]).
size mismatch for in_conv.bias: copying a param with shape torch.Size([8]) from checkpoint, the shape in current model is torch.Size([16]).
size mismatch for down_convs.0.conv.0.weight: copying a param with shape torch.Size([16, 8, 3, 3]) from checkpoint, the shape in current model is torch.Size([32, 16, 3, 3]).
size mismatch for down_convs.0.conv.0.bias: copying a param with shape torch.Size([16]) from checkpoint, the shape in current model is torch.Size([32]).
size mismatch for down_convs.0.conv.3.weight: copying a param with shape torch.Size([16, 16, 3, 3]) from checkpoint, the shape in current model is torch.Size([32, 32, 3, 3]).
size mismatch for down_convs.0.conv.3.bias: copying a param with shape torch.Size([16]) from checkpoint, the shape in current model is torch.Size([32]).
size mismatch for down_convs.1.conv.0.weight: copying a param with shape torch.Size([32, 16, 3, 3]) from checkpoint, the shape in current model is torch.Size([64, 32, 3, 3]).
size mismatch for down_convs.1.conv.0.bias: copying a param with shape torch.Size([32]) from checkpoint, the shape in current model is torch.Size([64]).
size mismatch for down_convs.1.conv.3.weight: copying a param with shape torch.Size([32, 32, 3, 3]) from checkpoint, the shape in current model is torch.Size([64, 64, 3, 3]).
size mismatch for down_convs.1.conv.3.bias: copying a param with shape torch.Size([32]) from checkpoint, the shape in current model is torch.Size([64]).
size mismatch for down_convs.2.conv.0.weight: copying a param with shape torch.Size([64, 32, 3, 3]) from checkpoint, the shape in current model is torch.Size([128, 64, 3, 3]).
size mismatch for down_convs.2.conv.0.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for down_convs.2.conv.3.weight: copying a param with shape torch.Size([64, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([128, 128, 3, 3]).
size mismatch for down_convs.2.conv.3.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for down_convs.3.conv.0.weight: copying a param with shape torch.Size([128, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([256, 128, 3, 3]).
size mismatch for down_convs.3.conv.0.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for down_convs.3.conv.3.weight: copying a param with shape torch.Size([128, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([256, 256, 3, 3]).
size mismatch for down_convs.3.conv.3.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for up_samples.0.conv.0.weight: copying a param with shape torch.Size([16, 32, 3, 3]) from checkpoint, the shape in current model is torch.Size([32, 64, 3, 3]).
size mismatch for up_samples.0.conv.0.bias: copying a param with shape torch.Size([16]) from checkpoint, the shape in current model is torch.Size([32]).
size mismatch for up_samples.1.conv.0.weight: copying a param with shape torch.Size([32, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([64, 128, 3, 3]).
size mismatch for up_samples.1.conv.0.bias: copying a param with shape torch.Size([32]) from checkpoint, the shape in current model is torch.Size([64]).
size mismatch for up_samples.2.conv.0.weight: copying a param with shape torch.Size([64, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([128, 256, 3, 3]).
size mismatch for up_samples.2.conv.0.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for up_samples.3.conv.0.weight: copying a param with shape torch.Size([128, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([256, 512, 3, 3]).
size mismatch for up_samples.3.conv.0.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for up_convs.0.conv.0.weight: copying a param with shape torch.Size([16, 32, 3, 3]) from checkpoint, the shape in current model is torch.Size([32, 64, 3, 3]).
size mismatch for up_convs.0.conv.0.bias: copying a param with shape torch.Size([16]) from checkpoint, the shape in current model is torch.Size([32]).
size mismatch for up_convs.0.conv.3.weight: copying a param with shape torch.Size([16, 16, 3, 3]) from checkpoint, the shape in current model is torch.Size([32, 32, 3, 3]).
size mismatch for up_convs.0.conv.3.bias: copying a param with shape torch.Size([16]) from checkpoint, the shape in current model is torch.Size([32]).
size mismatch for up_convs.1.conv.0.weight: copying a param with shape torch.Size([32, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([64, 128, 3, 3]).
size mismatch for up_convs.1.conv.0.bias: copying a param with shape torch.Size([32]) from checkpoint, the shape in current model is torch.Size([64]).
size mismatch for up_convs.1.conv.3.weight: copying a param with shape torch.Size([32, 32, 3, 3]) from checkpoint, the shape in current model is torch.Size([64, 64, 3, 3]).
size mismatch for up_convs.1.conv.3.bias: copying a param with shape torch.Size([32]) from checkpoint, the shape in current model is torch.Size([64]).
size mismatch for up_convs.2.conv.0.weight: copying a param with shape torch.Size([64, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([128, 256, 3, 3]).
size mismatch for up_convs.2.conv.0.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for up_convs.2.conv.3.weight: copying a param with shape torch.Size([64, 64, 3, 3]) from checkpoint, the shape in current model is torch.Size([128, 128, 3, 3]).
size mismatch for up_convs.2.conv.3.bias: copying a param with shape torch.Size([64]) from checkpoint, the shape in current model is torch.Size([128]).
size mismatch for up_convs.3.conv.0.weight: copying a param with shape torch.Size([128, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([256, 512, 3, 3]).
size mismatch for up_convs.3.conv.0.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for up_convs.3.conv.3.weight: copying a param with shape torch.Size([128, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([256, 256, 3, 3]).
size mismatch for up_convs.3.conv.3.bias: copying a param with shape torch.Size([128]) from checkpoint, the shape in current model is torch.Size([256]).
size mismatch for bottleneck_conv.conv.0.weight: copying a param with shape torch.Size([256, 128, 3, 3]) from checkpoint, the shape in current model is torch.Size([512, 256, 3, 3]).
size mismatch for bottleneck_conv.conv.0.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for bottleneck_conv.conv.3.weight: copying a param with shape torch.Size([256, 256, 3, 3]) from checkpoint, the shape in current model is torch.Size([512, 512, 3, 3]).
size mismatch for bottleneck_conv.conv.3.bias: copying a param with shape torch.Size([256]) from checkpoint, the shape in current model is torch.Size([512]).
size mismatch for out_conv.0.weight: copying a param with shape torch.Size([8, 16, 3, 3]) from checkpoint, the shape in current model is torch.Size([16, 32, 3, 3]).
size mismatch for out_conv.0.bias: copying a param with shape torch.Size([8]) from checkpoint, the shape in current model is torch.Size([16]).
size mismatch for out_conv.2.weight: copying a param with shape torch.Size([1, 8, 3, 3]) from checkpoint, the shape in current model is torch.Size([1, 16, 3, 3]).

[j-tseng]

Getting a similar error using the provided singularity image under section 3.1 of the README. I'm guessing the v1.0.9 beta-dimension bug fix addresses this - any chance you'll update the singularity image too? Thanks and eager to try this out!

[lianruizuo]

Thanks for the comments. We have fixed the dimension issue in v1.0.9 and provided an updated link for v1.0.9 singularity.