update

pages/2_batch_evaluation.py

@@ -0,0 +1,66 @@
+import pandas as pd
+import streamlit as st
+from util.evaluator import evaluator, write_evaluation_commentary
+import os
+
+# Predefined examples
+examples = {
+    'good': {
+        'question': "What causes rainbows to appear in the sky?",
+        'explanation': "Rainbows appear when sunlight is refracted, dispersed, and reflected inside water droplets in the atmosphere, resulting in a spectrum of light appearing in the sky."
+    },
+    'bad': {
+        'question': "What causes rainbows to appear in the sky?",
+        'explanation': "Rainbows happen because light in the sky gets mixed up and sometimes shows colors when it's raining or when there is water around."
+    }
+}
+
+# Function to check password
+def check_password():
+    password_input = st.sidebar.text_input("Enter Password:", type="password")
+    submit_button = st.sidebar.button("Submit")
+    if submit_button:
+        if password_input == os.getenv('PASSWORD'):
+            st.session_state['password_verified'] = True
+            st.experimental_rerun()
+        else:
+            st.sidebar.error("Incorrect Password, please try again.")
+
+def batch_evaluate(uploaded_file):
+    df = pd.read_csv(uploaded_file)
+    eval = evaluator(model_name='gpt4-1106')  # Assuming model name is fixed for simplicity
+    results = []
+
+    for _, row in df.iterrows():
+        question = row['question']
+        explanation = row['explanation']
+        scores = eval(question, explanation)
+        commentary = write_evaluation_commentary(scores)
+        results.append({**{'Question': question, 'Explanation': explanation}, **commentary})
+
+    result_df = pd.DataFrame(results)
+    return result_df
+
+# Title of the application
+st.title('Natural Language Explanation Demo')
+
+if 'password_verified' not in st.session_state or not st.session_state['password_verified']:
+    check_password()
+else:
+    st.sidebar.success("Password Verified. Proceed with the demo.")
+    st.header("Batch Evaluation of Questions and Explanations")
+    uploaded_file = st.file_uploader("Upload CSV file with columns 'question' and 'explanation'", type='csv')
+
+    if uploaded_file is not None:
+        if st.button('Evaluate Explanations'):
+            result_df = batch_evaluate(uploaded_file)
+            st.write('### Evaluated Results')
+            st.dataframe(result_df)
+
+            csv = result_df.to_csv(index=False)
+            st.download_button(
+                label="Download evaluation results as CSV",
+                data=csv,
+                file_name='evaluated_results.csv',
+                mime='text/csv'
+            )