update

pages/1_Generation.py

@@ -9,15 +9,19 @@ from util.analysis import statistical_tests, result_evaluation
 st.title('JobFair: A Benchmark for Fairness in LLM Employment Decision')
 st.sidebar.title('Model Settings')
 
+
 # Define a function to manage state initialization
 def initialize_state():
     keys = ["model_submitted", "api_key", "endpoint_url", "deployment_name", "temperature", "max_tokens",
-            "data_processed", "group_name","occupation", "privilege_label", "protect_label", "num_run", "uploaded_file"]
-    defaults = [False, "", "https://safeguard-monitor.openai.azure.com/", "gpt35-1106", 0.5, 150, False,"Gender", "Programmer", "Male", "Female", 1, None]
+            "data_processed", "group_name", "occupation", "privilege_label", "protect_label", "num_run",
+            "uploaded_file"]
+    defaults = [False, "", "https://safeguard-monitor.openai.azure.com/", "gpt35-1106", 0.5, 150, False, "Gender",
+                "Programmer", "Male", "Female", 1, None]
     for key, default in zip(keys, defaults):
         if key not in st.session_state:
             st.session_state[key] = default
 
+
 initialize_state()
 
 # Model selection and configuration
@@ -62,17 +66,20 @@ if st.session_state.model_submitted:
         if st.button('Process Data') and not st.session_state.data_processed:
             # Initialize the correct agent based on model type
             if model_type == 'AzureAgent':
-                agent = AzureAgent(st.session_state.api_key, st.session_state.endpoint_url, st.session_state.deployment_name)
+                agent = AzureAgent(st.session_state.api_key, st.session_state.endpoint_url,
+                                   st.session_state.deployment_name)
             else:
-                agent = GPTAgent(st.session_state.api_key, st.session_state.endpoint_url, st.session_state.deployment_name, api_version)
+                agent = GPTAgent(st.session_state.api_key, st.session_state.endpoint_url,
+                                 st.session_state.deployment_name, api_version)
 
             # Process data and display results
             with st.spinner('Processing data...'):
                 parameters = {"temperature": st.session_state.temperature, "max_tokens": st.session_state.max_tokens}
-                df = process_scores(df, st.session_state.num_run, parameters, st.session_state.privilege_label, st.session_state.protect_label, agent, st.session_state.group_name, st.session_state.occupation)
+                df = process_scores(df, st.session_state.num_run, parameters, st.session_state.privilege_label,
+                                    st.session_state.protect_label, agent, st.session_state.group_name,
+                                    st.session_state.occupation)
                 st.session_state.data_processed = True  # Mark as processed
 
-
             st.write('Processed Data:', df)
 
             # Allow downloading of the evaluation results
@@ -83,11 +90,6 @@ if st.session_state.model_submitted:
                 mime='text/csv',
             )
 
-
-
-
-
-
         if st.button("Reset Experiment Settings"):
             st.session_state.occupation = "Programmer"
             st.session_state.group_name = "Gender"
@@ -95,4 +97,4 @@ if st.session_state.model_submitted:
             st.session_state.protect_label = "Female"
             st.session_state.num_run = 1
             st.session_state.data_processed = False
-            st.session_state.uploaded_file = None
+            st.session_state.uploaded_file = None

pages/2_Evaluation.py

@@ -25,6 +25,7 @@ def app():
         if st.button('Evaluate Data'):
             with st.spinner('Evaluating data...'):
                 test_results = statistical_tests(df)
+                st.write('Test Results:', test_results)
                 evaluation_results = result_evaluation(test_results)
                 st.write('Evaluation Results:', evaluation_results)