Fitting process ending prematurely due to precision loss #96
Description
Hello!
i'm having issues fitting a potencial for a test dataset of 50 Cu structures, the fitting process should reach 2000 iterations before stopping but it ends prematurely only reaching up to 850 iterations and giving the same message,
my current input file fitting specifications are the following,
i've also tried with smaller gtols (like 1e-10) in case it made any difference, it didn't in all the cases i've tried i haven't been able to surpass the 850 iterations, i even tried with expanding my dataset to 150 structures but then it only reached 133 iterations before stopping.
I'm also curious if the "status=2" part of the error message has anything to do with why it's stopping early and maybe lead to a possible solution but i'm not sure.
Another thing i've also tried is using the L-BFGS-B optimizer since i understand it's made for bigger structures, with this i do surpass the 850 iterations threshold but even upping the iterations to 100000 (it only reached 15000) it's not capable of minimizing the RMSE energy below 700 meV/at so it doesn't work either.
If anybody knows what to do to up the precision in order for the potential to fully train i'd be immensely grateful if you helped me,
Kind regards,
Nick