Our team is currently looking to set up an evaluation for our modeling objective.
Currently we found that there’s no multi class library in the model objective. Do we need to write a library for it to get evaluation for a multi class classification? If we want to observe the evaluation of different versions of a model, what is the best method? Should we use model experiments, logs, save score and best param when training model in a dataset or something else?
That’s correct! A custom evaluator library can indeed be made to perform multi-class classification. Evaluation results can be built and viewed in the evaluations dashboard of your modeling objective, and multiple model versions, once submitted to the objective, can be chosen to compare evaluation results. Hope that helps!
1 Like
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.