search

bowang-lab / U-Mamba

U-Mamba: Enhancing Long-range Dependency for Biomedical Image Segmentation
https://arxiv.org/abs/2401.04722
Apache License 2.0
688 stars 64 forks source link

Argument passed to at() was not in the map. #22

Closed Overflowu7 closed 7 months ago

Overflowu7 commented 8 months ago

I'm having problems replacing Umaba's encoder and decoder with nnUNet's original decoder:

Argument passed to at() was not in the map.

And when this problem occurs, the pseudo-loss stays at 0. Have you seen this problem before? Looking forward to your reply!

JunMa11 commented 7 months ago

Hi @Overflowu7 ,

sorry for the late reply. We didn't try this. Could you please share more details?

Overflowu7 commented 7 months ago

thanks you for your reply! I will give the infomation. But this is a warning, not an error. Epoch 0 Current learning rate: 0.01 using pin_memory on device 1 /home/wu/wyc/nnUNet/nnunetv2/net/mamba/UMambaEnc.py:46: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! assert C == self.dim /home/wu/.conda/envs/nnUNet/lib/python3.9/site-packages/einops/einops.py:316: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! known: Set[str] = {axis for axis in composite_axis if axis_name2known_length[axis] != _unknown_axis_length} /home/wu/.conda/envs/nnUNet/lib/python3.9/site-packages/einops/einops.py:317: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs! unknown: Set[str] = {axis for axis in composite_axis if axis_name2known_length[axis] == _unknown_axis_length} 2024-04-02 21:22:12.595865: Unable to plot network architecture: 2024-04-02 21:22:12.596317: Argument passed to at() was not in the map. 2024-04-02 21:22:12.731663: 2024-04-02 21:22:12.731911: Epoch 0 2024-04-02 21:22:12.732270: Current learning rate: 0.01 using pin_memory on device 0

JunMa11 commented 7 months ago

How is the training going?