[BUG] SAM: Probabilities/Edge-Weights greater than 1 from method predict if not fullbatch

FenTechSolutions / CausalDiscoveryToolbox

Package for causal inference in graphs and in the pairwise settings. Tools for graph structure recovery and dependencies are included.

https://fentechsolutions.github.io/CausalDiscoveryToolbox/html/index.html

MIT License

1.12k stars 197 forks source link

[BUG] SAM: Probabilities/Edge-Weights greater than 1 from method predict if not fullbatch #164

Open Infinitif opened 6 months ago

Infinitif commented 6 months ago

Describe the bug If the batchsize is not -1 (i.e. not fullbatch), then the edge weights of the result graph obtained by calling sam.predict() are outside of the domain [0, 1]. This is consistent throughout all cdt-package-, python- and library versions and hardware configurations. Steps to reproduce: Just set the parameter _batchsize to something greater than 1 (eg. 2), _testepochs to sth. greater than 0 and call sam.predict().