Repository navigation
Make Sure Full Functionality is Available in C/C++ #476
Description
Activity
👍
I am currently working on expanding the C API to add support for graph definition. Not sure when it will be done, but it is one of our goals before 1.0.
Automatic differentiation will eventually be in, though closer to 1.0. Are you looking for C API support for use in another language or are you using C++ directly? We consider those separate APIs, but both should have something reasonable for defining graphs (though no shape inference or automatic differentiation, yet) in the next TensorFlow release (0.10.0) -- they are both checked in at head.
C++ graph-definition API update:
25ac3daC graph-definition API was a few commits, I'm having trouble tracking them all down.
Good to hear that! JavaCPP has no problems interacting with TensorFlow's C++ API directly, so using either API is fine in that case.
Andy will know more about the status of automatic differentiation.
Shape inference is now in C++ as is partial gradients. Work continues, but closing due to lack of recent activity. Please open a new issue referencing this one if it is still an issue. Thanks!
@aselle Do you mean to say that the C++ interface can now compute gradients and add them to the graph as in python? I don't see anything in the C interface to access this functionality. Is there a separate ticket to track that? Is there any rough timeline on when this functionality is expected?
Please make sure all the functionality is defined in the C++ core.