diff options
author | 2017-11-02 00:48:25 -0700 | |
---|---|---|
committer | 2017-11-02 00:51:20 -0700 | |
commit | df397f23b49004fa74934c4b6a28e89d988fb98e (patch) | |
tree | bfe403f7f5cf96fec474b261c4b6dcbcf8bc531d | |
parent | b611aec150977f7fc13735398eee5e4a4fcfd612 (diff) |
Internal change
PiperOrigin-RevId: 174293839
-rw-r--r-- | tensorflow/contrib/eager/README.OPENSOURCE.md | 15 |
1 files changed, 0 insertions, 15 deletions
diff --git a/tensorflow/contrib/eager/README.OPENSOURCE.md b/tensorflow/contrib/eager/README.OPENSOURCE.md deleted file mode 100644 index a4a3af08cf..0000000000 --- a/tensorflow/contrib/eager/README.OPENSOURCE.md +++ /dev/null @@ -1,15 +0,0 @@ -TensorFlow has many kernels for doing (deep) learning and data manipulation. -There are typically assembled into computational graphs which can run -efficiently in a variety of environments. - -We are exploring an alternative interaction, where kernels are invoked -immediately and call this "eager execution". We are hoping to retain the -benefits of graphs while improving usability with benefits like: - -- Immediate error messages and easier debugging -- Flexibility to use Python datastructures and control flow -- Reduced boilerplate - -Eager execution is under active development. -There are not many developer-facing materials yet, but stay tuned for updates -in this directory. |