0

I'm trying to assess how complicated a graph has to be to be useful. I'm interested in using RNN and LTSM, hence the question.

My impression is that a useful graph has many parts. I want to use the essential api in another language binding.

effbiae
  • 1,087
  • 1
  • 7
  • 22
  • 2
    Not an answer, but I experienced that the best way is to write the graph in Python, dump it to disk and load it using the c API ([this approach](https://stackoverflow.com/a/48893889/7443104) has been described before). I leave this as a comment as I am afraid a full answer is out-of-scope for such a small textbox below. – Patwie Apr 10 '18 at 06:56
  • @Patwie, I was browsing the Python source to lookup components of a graph and it all seemed so complicated - so many parts... Is that simply the nature of ML/Tensorflow? – effbiae Apr 10 '18 at 07:17

0 Answers0