combining two models

hello, I want to combine the model calculated on two different computers. The input attributes in these models are the same, only the values are different. how can I do that?

@mcstive usually parameter averaging works just fine. Sum the 2 parameters and divide by 2.

I wouldn’t suggest doing this outside the context of a dedicated training loop on the same dataset though (this is implemented by our parallelwrapper and spark integrations)