documentation

app.py

@@ -154,6 +154,11 @@ with st.form('model_test_prediction'):
             s3.metric(label ='Accuracy',value = acc, delta = '', delta_color = 'inverse')
   
 st.header("Comparison - Model Performance")
+st.write("""2 transformer models were finetuned and compared their performance on the test dataset. \\
+          - Bert uncased model (on Original & preprocessed text) \\
+          - Legal-Bert model (on Original & preprocessed text)\\
+        (Preprocessing steps were removal of numbers, symbols, stopwords followed by lemmatisation on tokens.)\\
+        The best performing model is Legal-BERT on original data. Please see the comparison below.""")
 met = pd.read_csv("model_comparison.csv")
 a1, a2 = st.columns((0.5,0.5))
 a1.subheader("Evaluation Metrics")