fire minimizer - NEB

Hello dear Lammps users,

Here I have some problems using NEB to find the energy barrier for some dislocation movement. I use "fire" minimizer and by giving 0 for both energy and force tolerances, I expect to minimization continue till the end for all the given number of iterations.
For the first stage of the NEB it does but for the climbing image part, the minimization stops much sooner and at the last the iteration the energy is all of sudden converged to value that makes no sense.
At first I tried to uses quickmin but the minimization takes so long that I started to think of other options.

Any suggestions?

Various things can go wrong when running a NEB calculation.
All of the minimizers can also decide to exit earlier than
the max # of iterations, if they aren't making progress. I.e.
they think they are done. Does the climbing stage print
out any info as to why the minimizer stopped? E.g. in
the per-replica log files.

Steve