Problems when training the model

[Mumuwei]

When training the model, I encountered the following problems：

in logger.py line 35
module.state_dict()['affine.weight'].copy_(checkpoint['kp_detector']['jacobian.weight'])
RuntimeError: The expanded size of the tensor (60) must match the existing size (40) at non-singleton dimension 0. Target sizes: [60, 35, 7, 7]. Tensor sizes: [40, 35, 7, 7]

[zhan-xu]

how did you solve the problem?

[zhan-xu]

i think we need to modify num_kp in first-order code to match num_segments in this code