[R-sig-ME] Do glmer and glmmadmb calculate log likelihood on the same scale ?

[Simon Chamaillé-Jammes]

Hello all, 
I would like to know if glmer and glmmadmb calculate log likelihood /
deviance on the same scale.

My glmer model end up with the 'false convergence warning'. However a
verbose output suggests that convergence actually goes not so badly. The
exact same model run without warning error using glmmadmb. The estimates
(fixed effects, variance components) are however widely different from
those estimated by glmer, and the logLik of the model is smaller but
negative with glmmadmb (-584.485) than with glmer (-424.4), but I'm
wondering if I can actually compare them. And if so, is this a sign I
could 'reasonably' trust the glmer output more than the glmmadmb ?

thanks

simon