How to save&load model under do not keep model in memory?

[oAzv]

When I keep model in memory, model can be saved and loaded.

BUT, with large training data, the RAM is run out quickly.

And, as same env, set keep memory set_keep_model_in_mem(False), the saved model only have header information without actually model content.

So, how to fix it?

[oAzv]

At present, I can find same matter in https://stackoverflow.com/questions/55020860/how-to-predict-using-a-gcforest-model-when-i-did-not-keep-the-model-in-memory, but does not resove this problem.