Different Topic Extractor (Parallel LDA) results despite seed

#1

Hi all,

I have a workflow that uses a LDA as it’s only random process and get different results for the same data set if I load a different data set inbetween.

Reproducibility is an important point for me, so I’d like to know how to fix this. I’m also interested in the inner workings of the node. Is there a hidden second random seed that is saved with the loaded dataset?

Thanks in advance,
m

0 Likes

#2

Hi @mfh -

Welcome to the forum, and sorry for the delayed response to your question.

Can you expand a little a bit on what you mean by “load[ing] a different data set in between?” If you have an example workflow that reliably reproduces the behavior you describe, could you perhaps post it for us to review?

0 Likes