Closed ClimbsRocks closed 8 years ago
then we let ensembling take it's pick.
it's generally good to have a bunch of un-correlated algos.
just make sure we don't spend 40 minutes training a larger version of them! skip the larger version part, but leave everything else the same.
skipped the longer training part, in addition to the larger classifier part.
we can definitely revisit the longTraining part later. I'm open to longTraining the crappy classifiers.
then we let ensembling take it's pick.
it's generally good to have a bunch of un-correlated algos.
just make sure we don't spend 40 minutes training a larger version of them! skip the larger version part, but leave everything else the same.