Show a random question from the moral scenarios evaluation

app.py

@@ -346,6 +346,14 @@ st.plotly_chart(fig)
 
 # Moral scenarios plots
 st.markdown("### Moral Scenarios Performance")
+def show_random_moral_scenarios_question():
+    moral_scenarios_data = pd.read_csv('moral_scenarios_questions.csv')
+    random_question = moral_scenarios_data.sample()
+    expander = st.expander("Show a random moral scenarios question")
+    expander.write(random_question['query'].values[0])
+
+show_random_moral_scenarios_question()
+
 st.write("""
          While smaller models can perform well at many tasks, the model size threshold for decent performance on moral scenarios is much higher.  
          There are no models with less than 13 billion parameters with performance much better than random chance. Further investigation into other capabilities that emerge at 13 billion parameters could help