From 7a7c1eb3fd85f46e79ed35d5a5468413efccfa2b Mon Sep 17 00:00:00 2001 From: Eugene Brevdo Date: Tue, 29 Nov 2016 09:38:37 -0800 Subject: [PATCH] Update release.md to reflect breaking changes to RNN variable names. Change: 140489119 --- RELEASE.md | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) diff --git a/RELEASE.md b/RELEASE.md index 939eee0f2d5eda..1b06bad4f5a6d0 100644 --- a/RELEASE.md +++ b/RELEASE.md @@ -16,7 +16,14 @@ BUS_ANY was used. indicating its preferred use in language bindings for TensorFlow. What was previously `TF_Session` has been renamed to `TF_DeprecatedSession`. * Renamed TF_Port to TF_Output in the C API. -* Removes RegisterShape from public API. Use C++ shape function registration instead. +* Removes RegisterShape from public API. Use C++ shape function registration + instead. +* Most RNN cells and RNN functions now use different variable scopes to be + consistent with layers (`tf.contrib.layers`). This means old checkpoints + written using this code will not load after this change without providing + `Saver` a list of variable renames. Examples of variable scope changes + include `RNN` -> `rnn` in `tf.nn.rnn`, `tf.nn.dynamic_rnn` and moving from + `Linear/Matrix` -> `weights` and `Linear/Bias` -> `biases` in most RNN cells. # Release 0.11.0