Can I combine multiple TensorFlow charts into one?

Question

I have two models trained with Tensorflow Python, exported to binaries with the name export1.metaand export2.meta. Both files will only generate one output when fed with input, for example, output1 and output2.

My question is whether it is possible to combine two graphs into one large graph so that it generates output1 and output2 in one execution.

Any comment would be helpful. Thanks in advance!

Answer 1

I kicked this with my local TF expert, and the short answer is no; For this TF does not have a built-in tool. However, you can write custom endpoint levels (input and output) with synchronizing operations from Python process control, so that they support parallel processing of each input and combine outputs.

Justification

I like how this can be used to get more precision with multiple functions where the functions have little or no correlation. For example, you can train two character recognition models: one to identify numbers, and the other to distinguish between left and right authors.

, , : "8" .

, , .

, , . , , / .