diff --git a/efficientdet/README.md b/efficientdet/README.md
index 7db802a0..59b0eff5 100644
--- a/efficientdet/README.md
+++ b/efficientdet/README.md
@@ -335,7 +335,7 @@ If you want to do inference for custom data, you can run
 
 You should check more details of runmode which is written in caption-4.
 
-## 9. Train on multi GPUs.
+## 9. Training on single node GPUs.
 
 Create a config file for the PASCAL VOC dataset called voc_config.yaml and put this in it.
 
diff --git a/efficientdet/tf2/README.md b/efficientdet/tf2/README.md
index a31d62b2..74fe8e69 100644
--- a/efficientdet/tf2/README.md
+++ b/efficientdet/tf2/README.md
@@ -260,11 +260,11 @@ Finetune needs to use --pretrained_ckpt.
 
 If you want to continue to train the model, simply re-run the above command because the `num_epochs` is a maximum number of epochs. For example, to reproduce the result of efficientdet-d0, set `--num_epochs=300` then run the command multiple times until the training is finished.
 
-## 9. Train on multi GPUs.
+## 9. Training on single node GPUs.
 
 Just add ```--strategy=gpus```
 
-## 10. Train on multi node GPUs.
+## 10. Training on multi node GPUs.
 Following scripts will start a training task with 2 nodes.
 
 Start Chief training node.