In xgboost 1.6+, they introduced a method for exporting the json version of classifiers.
We're currently on 1.7 ever since #27.
Their docs: https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Right now we use a serialized memory snapshot, but it would be relatively easy to switch that to their json "model + config" format. At least in the docs, they guarantee backwards compatibility.
In xgboost 1.6+, they introduced a method for exporting the json version of classifiers.
We're currently on 1.7 ever since #27.
Their docs: https://xgboost.readthedocs.io/en/latest/tutorials/saving_model.html
Right now we use a serialized memory snapshot, but it would be relatively easy to switch that to their json "model + config" format. At least in the docs, they guarantee backwards compatibility.