From 409c44d362803598cc3efa19d10412d64de0edea Mon Sep 17 00:00:00 2001
From: Mike J Innes <mike.j.innes@gmail.com>
Date: Tue, 4 Oct 2016 22:50:49 +0100
Subject: [PATCH] comments

---
 README.md | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 0769dd2c..b1dc1c20 100644
--- a/README.md
+++ b/README.md
@@ -86,9 +86,8 @@ Though further from the equations, this has the advantage of further reuse and c
 When it comes time to train the model, we have a number of options for tweaking its implementation, like the backend used, or batching and unrolling settings. In Flux this is as simple as calling some functions on the original model:
 
 ```julia
-model = unroll(model, 10)
-model = batch(model, 100)
-model = mxnet(model)
+model = unroll(model, 10) # Statically unroll the model
+model = tf(model) # Convert the model to TensorFlow
 
 Flux.train!(model, ...)
 ```