WARNING: Fix qeq/reax CG convergence failed after 200 iterations at 1 step and ERROR on proc 0: Non-numeric atom coords - simulation unstable

This is most probably initial configuration errror. There are some atoms very close to each other.
Try to build system again or you can try running at very low temperature first then minimize the system. Or you can degrease the charge accuracy first run couple of thousand steps and change back to original.