From 354aa119086c27d54d49bcad5536022008ac8dfa Mon Sep 17 00:00:00 2001
From: guru4elephant <35550832+guru4elephant@users.noreply.github.com>
Date: Wed, 21 Nov 2018 16:12:57 +0800
Subject: [PATCH] Update async_executor.md

---
 doc/fluid/design/async_executor/async_executor.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/doc/fluid/design/async_executor/async_executor.md b/doc/fluid/design/async_executor/async_executor.md
index d9baa9efa..ae57daec8 100644
--- a/doc/fluid/design/async_executor/async_executor.md
+++ b/doc/fluid/design/async_executor/async_executor.md
@@ -6,10 +6,10 @@ There are many deep learning applications that use sparse features as inputs, su
 ``` python
 def train_loop():
     filelist = ["testfile.data"] # filelist file to be handled
-    dataset = fluid.DataFeedDesc('data.prototxt') # this prototxt is a datafile description protobuf
-    dataset.set_batch_size(1)     # datafeed should be assigned a batch size
     data = fluid.layers.data(name="doc", shape=[1], dtype="int64", lod_level=1) # input text data
     label = fluid.layers.data(name="title", shape=[1], dtype="int64", lod_level=1) # label data
+    dataset = fluid.MultiSlotData('data.prototxt', data, label)
+    dataset.set_batch_size(128)
     avg_cost, acc, prediction = bow_net(data, label)
     sgd_optimizer = fluid.optimizer.Adagrad(learning_rate=0.002)
     opt_ops, weight_and_grad = sgd_optimizer.minimize(avg_cost)
-- 
GitLab