Thank you for your message. The figures show incomplete but let me try to answer your questions:
1) To increase the ability of the fitting tool to fit more precisely the data I recommend increasing the population size (by default 400) and decreasing the tolerance (by default 1e-11).
This also brings longer computation times, so another approach is run a standard fitting procedure and then play around with the values manually until you get a perfect fit (for example, you could play around with that trap -1 that does not seem to do anything until you fit that high temperature peak
2) The very beginning of the test always looks a bit different to the experiments. This is because of the resting time of the experiments and because the H content is never fully homogeneous in the sample. Numerical results typically show a much higher flux as they idealise the exposure step.
3) Yes, you can ignore traps whose trap density/binding energy after the fitting are tiny/negligible
I hope this helps
