diff --git a/docs/getting_started.rst b/docs/getting_started.rst
index d227b9554..5d1d2d697 100644
--- a/docs/getting_started.rst
+++ b/docs/getting_started.rst
@@ -90,7 +90,7 @@ well for the IWSLT 2014 dataset:
 
     > mkdir -p checkpoints/fconv
     > CUDA_VISIBLE_DEVICES=0 fairseq-train data-bin/iwslt14.tokenized.de-en \
-        --lr 0.25 --clip-norm 0.1 --dropout 0.2 --max-tokens 4000 \
+        --optimizer nag --lr 0.25 --clip-norm 0.1 --dropout 0.2 --max-tokens 4000 \
         --arch fconv_iwslt_de_en --save-dir checkpoints/fconv
 
 By default, :ref:`fairseq-train` will use all available GPUs on your machine. Use the