Topaz Validate generates a bunch of Topaz Train jobs varying a specific parameter - e.g. expected particle number.
However, when I look at the individual Topaz Train jobs run by Cross Validate, they all have the same value set for this parameter in the GUI - even though when I look at the log it is clear that the actual Topaz command has been executed with the correct value. This is a little confusing.
Hey @olibclarke –
Thanks for reporting. This is noted. FYI: we recommend using workflows instead of the cross validate job now.
Thanks Valentin! Out of interest, why would you recommend workflows over topaz cross validate for this purpose? Cross validate can auto generate the jobs with parameters varying across a specified range, and then run a final topaz train job for the best set of params - with Workflows, wouldn’t I need to create a custom workflow for each different use case?
Ah sorry about that – I was a little confused, you are totally right, you won’t be able to replicate the cross validation behaviour via a straightforward workflow. In any case, thanks for reporting. We’ve noted the discrepancy.