search

sphinxteam / Boltzmann.jl

Restricted Boltzmann Machines in Julia
Other
15 stars 4 forks source link

Bug (some bugs listed recently) #21

Open AurelienDecelle opened 7 years ago

AurelienDecelle commented 7 years ago

First, the constructor of the RBM is not working : Normal do not take tuple in argument anymore. Second : the sampling for gaussian variables is using the bernoulli function for that Third : the fit function do not work correctly for gaussian data because of the @assert

Something else : the gibbs functino is completely obsolete I guess (and buggy).

A.

krzakala commented 7 years ago

Arg, Julia is evolving too fast

eric-tramel commented 7 years ago

Hi @AurelienDecelle, Thanks for pointing out the problems in the current iteration. I want to plan on updating the package, taking into account the latest developments we've made in the latest GRBM paper we have on arXiv. Everything should generalize sufficiently well, and I think that in the meantime there are a number of things that we need to do in order to update this package to Julia 0.5.

I'll start taking a look at what can be done.