Umbrella sampling: starting frame high distance

GROMACS version: 2023
GROMACS modification: No
Dear Users,
I performed a pull protocol for subsequent umbrella sampling. After have got COM distances, I noticed that pullx plot report a large distance from the first frames (0.966; please, see also attached files). I was expecting to have a much lower initial values, say 0.2 to 0.5nm distance. Should I worry about that? Nevertheless, aftre US, I obtained the attached histogram of the sampled configurations, which shows a gap in the first portion of the plot. Is it due to the previous issue?
Thanks in advance
pullx_0.005_500kJ.xvg (101.3 KB)
pullf_0.005_500kJ.xvg (99.5 KB)
histo.xvg (112.5 KB)