Hi! I’m new to Tensorflow1.0. Now I have pretrained a model and I want to train another model with the output of the pretrained model (The pretrained model will not be updated during training). The two models share some of the inputs. I know to do this I have to load the pretrained model and build the training in different graphs. But what about the training input? Since the two model share some of the inputs, do I have to build the training inputs in both graphs or there is anther ways to avoid that? Thanks!
HI! I’m new to Tensorflow 1.0. Now I have a pretrained model and I want to train another model with the output of the pretrained model (the pretrained model will not be updated during the training). The two model share some common inputs. I know I need to load the pretrained model and build the training model in different graphs. But how about the inputs? Since the two graphs share some of the inputs, do I have to build the same inputs in different graphs? Or there are some other methods to avoid that? Thanks!!