This message was deleted.

yes! this one: https://eng.uber.com/introducing-ludwig/ i was wondering what the best format would be for using it with bento. I've seen that there is a tensorflow framework, but my colleague said that ludwig applies a series of preprocessing steps that could not be contemplated by saving the model directly as a tensorflow one

Hello, @Jessica de Paula Liporace. You could do your own Ludwig Model Artifact. It's not hard. Maybe just copy the PytorchModelArtifact class and modify the methods. Or follow the bentoml docs instructions for custom model artifacts: https://docs.bentoml.org/en/latest/guides/custom_artifact.html

thank you! I'll give it a try