Modification on baseline ranker, how we get the max value on a list of ranks to normalize the ranks. The previous version calculated the max on each iteration which took a couple of hours when the dataset was big enough. With this change, it takes a couple of seconds to normalize the ranks.
Test units associated to the ranker
Changes in versions of the requirements in dev, main, build, and test-management that were required for the actions jobs to run properly on git
Adding to predict-list the cohort_config_from_label_config for the cases where the cohort configuration is defined within the label config.
Changes
dev
,main
,build
, andtest-management
that were required for the actions jobs to run properly on gitpredict-list
thecohort_config_from_label_config
for the cases where the cohort configuration is defined within the label config.