1. The most common problem is whether two settings correspond with each other, including:

    • The order of elements in type_map and mass_map and fp_pp_files.

    • Size of init_data_sys and init_batch_size.

    • Size of sys_configs and sys_batch_size.

    • Size of sel_a and actual types of atoms in your system.

    • Index of sys_configs and sys_idx.

  2. Please verify the directories of sys_configs. If there isn’t any POSCAR for 01.model_devi in one iteration, it may happen that you write the false path of sys_configs. Note that init_data_sys is a list, while sys_configs should be a two-dimensional list. The first dimension corresponds to sys_idx, and the second level are some poscars under each group. Refer to the sample file.

  3. Correct format of JSON file.

  4. The frames of one system should be larger than batch_size and numb_test in default_training_param. It happens that one iteration adds only a few structures and causes error in next iteration’s training. In this condition, you may let fp_task_min be larger than numb_test.

  5. If you found the dpgen with the same version on two machines behaves differently, you may have modified the code in one of them.