Hi,
After trying to run the code for CelebA (with ResNet18), I get the following results:
- valid_attrwise_accs_b
tensor([[0.9578, 0.9966], [0.8382, 0.2747]])
- valid_attrwise_accs_d
tensor([[0.0300, 0.0205], [0.6841, 0.9231]])
Could you please specify the exact way to replicate the results in the paper?
Thank you!