From 48e3b625410d4854296c6ddc8a2b102d25e483ba Mon Sep 17 00:00:00 2001
From: Eugene Brevdo <ebrevdo@google.com>
Date: Wed, 30 Aug 2017 09:27:00 -0700
Subject: Delete tf.contrib.training.python_input.

It has been replaced by tf.contrib.data.Dataset.from_generator.

PiperOrigin-RevId: 167004190
---
 RELEASE.md | 3 +++
 1 file changed, 3 insertions(+)

(limited to 'RELEASE.md')

diff --git a/RELEASE.md b/RELEASE.md
index ec66e555ad..d120f068ca 100644
--- a/RELEASE.md
+++ b/RELEASE.md
@@ -9,6 +9,9 @@
   for LSTMs and stacked LSTMs.  This bug fix follows recommendations from
   published literature, but is a behavioral change.  State dropout behavior
   may be customized via the new `dropout_state_filter_visitor` argument.
+* Removed `tf.contrib.training.python_input`.  The same behavior, in a more
+  flexible and reproducible package, is available via the new
+  `tf.contrib.data.Dataset.from_generator` method!
 
 # Release 1.3.0
 
-- 
cgit v1.2.3