Given transposed=1, weight of size [768, 128, 3, 3, 3], expected input[4, 512, 3, 3, 3] to have 768 channels, but got 512 channels instead

reinforcement-learning
1

Hello, I am having a problem with a neural network, I am a beginner and so even if I found similar topics I don’t understand where my problem is. The network I am using is the following:
GitHub - kbressem/prostate158 (is a model using a UNet structure)
and my problem is that after training the model, when I try to evaluate the model on the test dataset using the code that is suggested in the README, I get the following error:

RuntimeError: Given transposed=1, weight of size [768, 128, 3, 3, 3], expected input[4, 512, 3, 3, 3] to have 768 channels, but got 512 channels instead

I really don’t know how to fix this and I would really appreciate your help.

image
image1472×828 60.2 KB

2

Could you post the arguments you are using to create the UNet from MONAI, please?

3

I am using the code provided by the github page (shown in the picture), where the anatomy.yaml file is as follows:

code
code1445×752 44 KB