Loading model from checkpoint after error in training

did you get an answer for this?