Update pages/3_QA.py

pages/3_QA.py

@@ -26,6 +26,19 @@ st.markdown("""
     <li>Importantly the fine tuned model used for Q & A was DistilledBert squad. </li>
 </ul>
 </div>""",unsafe_allow_html=True)
+
+st.markdown("""<div class="feature_text"> Reasons for choosing DistilledBert squad,
+</div>""",unsafe_allow_html=True)
+st.markdown("""
+<div class="feature_text">
+<ul>
+    <li> Firstly the DistilledBert model retained almost the same accuracy as the bigger BERT dataset </li>
+    <li> Whilst it holds almost the same accuracy the size of the model is siginificantly smaller meaning it is around 60% faster </li>
+    <li> Due to being a smaller model it is very strong for a smaller to medium size >200mb csv files that are being classified in this app </li>
+    <li> The model is also optimized to be bi-directional so it checks the answer key from both the starting and ending point </li>
+</ul>
+</div>""",unsafe_allow_html=True)
+
 #creating a loop to see if a csv file is there
 if "uploaded_content" not in st.session_state:
     st.warning("Please upload and classify a CSV on the Classification page first.")