Update README.md

Change Colab links

README.md

@@ -32,7 +32,7 @@ The scheduled learning rate is dampened by the multiplication of the warmup fact
 <p align="center"><img src="https://github.com/Tony-Y/pytorch_warmup/raw/master/examples/emnist/figs/learning_rate.png" alt="Learning rate" width="400"/></p>
 
 #### Approach 1
-[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/Tony-Y/colab-notebooks/blob/master/PyTorch_Warmup_Approach1.ipynb)
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/Tony-Y/colab-notebooks/blob/master/PyTorch_Warmup_Approach1_chaining.ipynb)
 
 When the learning rate schedule uses the global iteration number, the untuned linear warmup can be used as follows:
 
@@ -69,7 +69,7 @@ for epoch in range(1,num_epochs+1):
 ```
 
 #### Approach 2
-[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/Tony-Y/colab-notebooks/blob/master/PyTorch_Warmup_Approach2.ipynb)
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/Tony-Y/colab-notebooks/blob/master/PyTorch_Warmup_Approach2_chaining.ipynb)
 
 When the learning rate schedule uses the epoch number, the warmup schedule can be used as follows: