• Removed parameter rng from the ML and MAP trainers (issue #25)
  • Changed debug flags in the the EM training (issue #26)
  • Changed dependency with bob.math (>= 3.0.0) (updated linsolve function)