Dependencies and steps through which Tensorflow (TF) gets installed on your development machine.
As of Dec 2019, the instructions to install Tensorflow (TF) are here instructions and one should run:
$ pip install tensorflow
To learn more about pip see the (pip note)[https://winding-lines.github.io/notes/pip/,
For python 3.7 TF is provided through 3 files, one per operating system:
The manylinux2010
is platform tag which allows binary wheels to be installed on compatible linux libraries. This is probably the reason why TF requires pip version 19, as per the changelog this platform support first appeared in this version.
The macos wheel contains the following files:
Ext | Files | Largest file | Size [kB] |
---|---|---|---|
.py | 1510 | tensorflow_core/python/ops/gen_nn_ops.py | 738.56 |
.h | 2219 | tensorflow_core/include/Eigen/src/misc/lapacke.h | 1033.56 |
.dylib | 3 | tensorflow_core/libtensorflow_framework.2.0.0.dylib | 27937.71 |
.so | 7 | tensorflow_core/python/_pywrap_tensorflow_internal.so | 321594.8 |
.inc | 25 | tensorflow_core/include/google/protobuf/test_util.inc | 123.58 |
.c | 75 | tensorflow_core/include/external/boringssl/src/third_party/fiat/p256.c | 80.71 |
These files are split amongst the following top-level folders:
The wheel contains the following pre-compiled libraries:
It has the following outside dependencies:
Tensorflow internals: https://github.com/horance-liu/tensorflow-internals