Seq2Seq model loss not decreasing

I completed the translation tutorial on the Pytorch website and was trying to adapt it to a different dataset. But the issue is that whenever I try to train the model, the loss is stuck in a single range between 5.9 and 6.9. It is not decreasing and remains in the same range. What can be the possible issue?
Link to Google Colab notebook- https://colab.research.google.com/drive/1iVg9-XxFlA-3aXBEReeSbZD9gF_7jsys