robustML/advertrain/dependencies/fire.py

• 1 F841: local variable 'batch_size' is assigned to but never used
  • 97 batch_size = len(x_natural)
• 2 E261: at least two spaces before inline comment
  • 110 loss_kl = (torch.acos(sqroot_prod - 1e-7) ** 2).mean(0) # Minus eps to prevent gradient explosion near 1 (https://github.com/pytorch/pytorch/issues/8069)
  • 145 loss_robust = (torch.acos(sqroot_prod - 1e-7) ** 2).mean(0) # Minus eps to prevent gradient explosion near 1 (https://github.com/pytorch/pytorch/issues/8069)
• 2 E501: line too long (177 > 120 characters) (and 1 similar)

  • Line too long (177 > 120 characters):

    110 loss_kl = (torch.acos(sqroot_prod - 1e-7) ** 2).mean(0) # Minus eps to prevent gradient explosion near 1 (https://github.com/pytorch/pytorch/issues/8069)

  • Line too long (165 > 120 characters):

    145 loss_robust = (torch.acos(sqroot_prod - 1e-7) ** 2).mean(0) # Minus eps to prevent gradient explosion near 1 (https://github.com/pytorch/pytorch/issues/8069)
• 1 W291: trailing whitespace
  • 168 based on the model's predictions for the noisy data.