I can't reproduce the Cu-Mg example

[BenedicteM]

Hi,
I am willing to use ESPEI for optimizing binary systems.
So first of all, I am trying to reproduce the Cu-Mg example.
I am using the development version of ESPEI.
espei 0.8.7
pycalphad 0.10.0

After the parameter generation step, ESPEI has generated 15 parameters instead of the 18 mentioned in the tutorial, my figure is not the same but it looks reasonable:

After the MCMC optimization step, the liquidus temperatures are improved but not the solubilities of the fcc and Laves phases.

Do you know what I missed ?
Thanks,
Bénédicte