Dear Pallavi Agrawal

Potential energy in this case should be negative, probably. Take into account that starting configuration can be (and in practice always is) very very far from an 'equilibrium' potential energy. That's why you do energy minimization.

Change scale, it is likely that the energy dropped by several orders of magnitude and you can't see it on a 10^15 y axis. Open the potential.xvg file and look at the end values in it. Set the y scale around those values, you should be able to appreciate better the decrease and if it has converged to some acceptable asymptotic value.

Hi Pallavi! I do not know what you are worried about, but I can tell you that you do energy minimization in order to avoid overlaps of atoms/molecules and get more reasonable structure for equilibration and production run. In general, if you will be able to start the equilibration without a crash of your simulation then you do not need to care at all about what was during the minimization. After all highly likely you won't reach any global minima and all algorithms (steepest descent and conjugate gradients) have their own limitations. If you have some strange interest to sit and minimize more you may play with values for tolerance (emtol) and the step (emstep) as well as change constraints from h-bonds to all-bonds. Such manipulations can help you to get better structures for your equilibration.