Update README.md

README.md

@@ -17,7 +17,7 @@ It's using Paddle to handle the training and other things. Not PyTorch or Tensor
 There's the result I get form this model:
 
 - You can use CPU to run the model! But GPU 10x faster then CPU 🚀.
-- CPU (about 1200ms/per  ) | GPU (about 200ms/per 🚀  )
+- CPU (about 300ms/per  ) | GPU (about 90ms/per 🚀  ) V2-10 Model
 - You can add some change easier passing some params.
 
 ## 📕 Using example is here