Python is the primary language in which TensorFlow models are typically developed and trained. TensorFlow does have bindings for other programming languages. These bindings have the low-level primitives that are required to build a more complete API, however, lack much of the higher-level API richness of the Python bindings, particularly for defining the model structure.

This gist demonstrates taking a model (a TensorFlow graph) created by a Python program and running the training loop in a C program.

All packages, except for Tini have been added to termux-root. To install them, simply pkg install root-repo && pkg install docker. This will install the whole docker suite, left only Tini to be compiled manually.