SynapseML
SynapseML copied to clipboard
How to track lightgbm eval performance during and post training?
One can specify the eval data and eval metric for lightgbm via the validationIndicatorCol
and metric
parameters. How can one:
- See the iteration by iteration evaluation metric evolution like one can with non-spark-lightgbm?
- See the evaluation evolution information (from point 1) post model training. At least something like
transformer.bestEvalScore(evalDataId, evalMetricId)
. Best would be something liketranformer.getEvalResults(evalDataId, evalMetricId)
. Of course, theevalDataId
andevalMetricId
need to have been assigned before the training started
AB#1984493
Tagging in the expert: @imatiach-msft