Merge branch 'master' of https://github.com/samjabrahams/tensorflow-o…

…n-raspberry-pi

Remove protobuf instructions, add OS details

Update README and GUIDE for 1.0.0

Merge pull request #66 from samjabrahams/0.12

Update to 12.1

Actually remove archive

Update filenames in README.md

Update to version 0.9.0

Remove '-mfpu=neon' GCC compiler option

The newer versions of TensorFlow use
[`Eigen::half`](https://eigen.tuxfamily.org/dox-devel/Half_8h_source.html)
to allow for 16-bit floating point operations. Unfortunately, these are
not compatible with the NEON instruction set. To keep things simple
(and functioning), the final `bazel build` step for creating the
TensorFlow wheel no longer has `--copt="-mfpu=neon"` included.

TensorFlow v0.8.0 full release

RPi3 TensorFlow 0.8 Release Candidate 0