Skip to content

v1.3.1

Compare
Choose a tag to compare
@amcadmus amcadmus released this 12 Jan 20:08
· 1718 commits to master since this release
4002aa5

Bug fixing:

  • Compulsory label requirement if the corresponding prefactor is set to non-zero. The current behavior is when then label is missing, the corresponding term does not appear in the loss function
  • Optional requirement of loss in dipole and polarizability training

Improvement:

  • Recommend consistent TensorFlow versions for python and C++ interfaces.