Update README.md

README.md

@@ -98,7 +98,10 @@ The model was pre-trained continuously on a single A10G GPU in an AWS instance f
 
 2. We have pre-trained our model with approx 16 GB of data, and testing Classification result on <a href='https://www.kaggle.com/datasets/ashokpant/nepali-news-dataset-large/data'>Nepali News Dataset (Large)</a> with a couple of Nepali transformer based Models available on Hugging Face,
 <br> Our models seem to do better than others with an accuracy of 0.58 on validation but,
-<br> It's seen that we still do not have enough data for generalization as Transformer models only perform well with large amounts of pre-trained data compared with Classical Sequential Models.
+<br> There could be two reasons for this:
+
+   - There is still room for improving the quality of the data.
+   - It's seen that we still do not have enough data for generalization as Transformer models only perform well with large amounts of pre-trained data compared with Classical Sequential Models.
 
 #### Authors: