Closed benwulfe closed 1 year ago
Gonna assign this to myself and dig deeper into this tomorrow.
ML working session suggests this might not be a priority, compared to improving the variational model, but working session for this has been pushed back to Friday.
This is covered by the higher level tracking bug. closing.
Very likely the radius chosen for the gaussian blur is not optimal and can be further tuned while testing. See the method
get_2d_gaussian
(xgboost/hosted_runtime/xgboost_hosted_runtime
,xgboost/main
).