There is currently not a default model evaluator for multi class classification models. There also does not seem to be support for having multiclass style outputs in models (lists of floats). This request is to support that in the default evaluators, to enable list outputs in prediction outputs, and to document implementation details for this model type.
This is a common modelling need and will support use cases to deploy the updates to Foundry ML.
Hey @devonyates, we are building an easier workflow for Model Evaluation rolling out in the next couple weeks. Happy to provide documentation when it is available.
1 Like