Why can't my training results converge?

In this stenosis2dsimplecase_0hb.py code, the results after training did not converge; the convergence of loss_continuity, loss_u, and loss_v is relatively poor. Is there any normal convergence? What could be the problem?