diff --git a/README.md b/README.md
index 49edb9e2df3e11f4df5451c0ce41ea781bdbe8f0..bce7f8e9fe1879732e854b795981b8af9523000c 100644
--- a/README.md
+++ b/README.md
@@ -35,12 +35,19 @@ then you can see in the terminel as follow:
 ![attentionocr](./test_img/md_img/attentionV2.png)
 there uses the decoderV2 model for decoder.
 
+
+# The previous version  
+
+**_git checkout AttentionOcrV1_**
+
+
 # Reference
 1. [crnn.pytorch](https://github.com/meijieru/crnn.pytorch)
 2. [Attention-OCR](https://github.com/da03/Attention-OCR)
 3. [Seq2Seq-PyTorch](https://github.com/MaximumEntropy/Seq2Seq-PyTorch)
 4. [caffe_ocr](https://github.com/senlinuc/caffe_ocr)
 
+
 # TO DO
 - [ ] change LSTM to Conv1D, it can greatly accelerate the inference
 - [ ] change the cnn bone model with inception net, densenet