Some questions about boundary conditions in the dataset.

[nnyyxx-bit]

I understand that the nopbc file is an identifier of the boundary conditions when the model is trained. The nopbc file does exist in my dataset, but I have two questions about this.

1. I used init_reaction to generate the initial data, I don't seem to have a way to set the boundary conditions through the reaction.json (I need pbc), so I indirectly modified the in.lmp file used to run Reaxff MD by modifying the source code to set the boundary conditions to pbc, but the nopbc file still exists in the resulting dataset.
2. In the 01.model_devi step, I set the boundary condition by setting "model_devi_nopbc" to "false". However, the nopbc file still exists in the resulting dataset.
   
   So how are the boundary conditions recognised when generating the dataset? How can I change it?