The negative slope is problematic. That will lead to instabilities during the solution.
I suggest you discard the stress-strain data below a strain value of 0.03 and fit a lower order curve to the remaining points. That should eliminate the negative slope.
You are navigating away from the AIS Discovery experience