Update app.py

app.py

@@ -94,7 +94,7 @@ def load_and_classify_csv_dataframe(file, text_field, event_model, threshold):
     return result_df, result_df, gr.update(choices=sorted(filters+extra_filters), 
                                                             value='All', 
                                                             label="Filter data by label", 
-                                                            visible=True)
+                                                            visible=True), gr.update(interactive=True), gr.update(interactive=True)
 
 
 def calculate_accuracy(flood_selections, fire_selections, none_selections, num_posts, text_field, data_df):
@@ -172,7 +172,6 @@ with gr.Blocks(fill_width=True) as demo:
     
     T_data_ss_state = gr.State(value=pd.DataFrame())
     
-    
     with gr.Tab("Event Type Classification"):
         gr.Markdown(
         """
@@ -213,75 +212,6 @@ with gr.Blocks(fill_width=True) as demo:
 
     
 
-    with gr.Tab("Event Type Classification Eval"):
-        gr.Markdown(
-        """
-        # T4.5 Relevance Classifier Demo 
-        This is a demo created to explore floods and wildfire classification in social media posts.\n
-        Usage:\n
-            - Upload .tsv or .csv data file (must contain a text column with social media posts).\n
-            - Next, type the name of the text column.\n
-            - Then, choose a BERT classifier model from the drop down.\n
-            - Finally, click the 'start prediction' buttton.\n
-        Evaluation:\n
-            - To evaluate the model's accuracy select the INCORRECT classifications using the checkboxes in front of each post.\n
-            - Then, click on the 'Calculate Accuracy' button.\n
-            - Then, click on the 'Download data as CSV' to get the classifications and evaluation data as a .csv file.
-        """)
-        with gr.Row():
-            with gr.Column(scale=4):
-                file_input = gr.File(label="Upload CSV or TSV File", file_types=['.tsv', '.csv'])
-                
-            with gr.Column(scale=6):
-                text_field = gr.Textbox(label="Text field name", value="tweet_text")
-                event_model = gr.Dropdown(event_models, value=event_models[0], label="Select classification model")
-                ETCE_predict_button = gr.Button("Start Prediction")
-        with gr.Accordion("Prediction threshold", open=False):        
-            threshold = gr.Slider(0, 1, value=0, step=0.01, label="Prediction threshold", show_label=False, 
-                              info="This value sets a threshold by which texts classified flood or fire are accepted, \
-                                  higher values makes the classifier stricter (CAUTION: A value of 1 will set all predictions as none)", interactive=True)
-        
-        with gr.Row(): 
-            with gr.Column():
-                gr.Markdown("""### Flood-related""")
-                flood_checkbox_output = gr.CheckboxGroup(label="Select ONLY incorrect classifications", interactive=True)
-                
-            with gr.Column():
-                gr.Markdown("""### Fire-related""")
-                fire_checkbox_output = gr.CheckboxGroup(label="Select ONLY incorrect classifications", interactive=True)
-                
-            with gr.Column():
-                gr.Markdown("""### None""")
-                none_checkbox_output = gr.CheckboxGroup(label="Select ONLY incorrect classifications", interactive=True) 
-
-        with gr.Row():
-            with gr.Column(scale=5):
-                gr.Markdown(r"""
-                Accuracy: is the model's ability to make correct predicitons.
-                It is the fraction of correct prediction out of the total predictions.
-                
-                $$
-                \text{Accuracy} = \frac{\text{Correct predictions}}{\text{All predictions}} * 100
-                $$
-                
-                Model Confidence: is the mean probabilty of each case 
-                belonging to their assigned classes. A value of 1 is best.
-                """, latex_delimiters=[{ "left": "$$", "right": "$$", "display": True }])
-                gr.Markdown("\n\n\n")
-                model_confidence = gr.Number(label="Model Confidence")
-                
-            with gr.Column(scale=5):
-                correct = gr.Number(label="Number of correct classifications")
-                incorrect = gr.Number(label="Number of incorrect classifications")
-                accuracy = gr.Number(label="Model Accuracy (%)")
-
-        ETCE_accuracy_button = gr.Button("Calculate Accuracy")
-        download_csv = gr.DownloadButton(visible=False)
-        num_posts = gr.Number(visible=False)
-        data = gr.DataFrame(visible=False) 
-        data_eval = gr.DataFrame(visible=False)
-        
-
     qa_tab = gr.Tab("Question Answering")
     with qa_tab:
         gr.Markdown(
@@ -313,7 +243,7 @@ with gr.Blocks(fill_width=True) as demo:
         QA_run_button = gr.Button("Start QA", interactive=False)
         hsummary = gr.Textbox(label="Summary")
         
-        qa_df = gr.DataFrame()
+        qa_df = gr.DataFrame(visible=False)
 
 
     with gr.Tab("Single Text Classification"):
@@ -346,8 +276,75 @@ with gr.Blocks(fill_width=True) as demo:
                 classification_score = gr.Number(label="Classification Score")
  
         
-
+    # with gr.Tab("Event Type Classification Eval"):
+    #     gr.Markdown(
+    #     """
+    #     # T4.5 Relevance Classifier Demo 
+    #     This is a demo created to explore floods and wildfire classification in social media posts.\n
+    #     Usage:\n
+    #         - Upload .tsv or .csv data file (must contain a text column with social media posts).\n
+    #         - Next, type the name of the text column.\n
+    #         - Then, choose a BERT classifier model from the drop down.\n
+    #         - Finally, click the 'start prediction' buttton.\n
+    #     Evaluation:\n
+    #         - To evaluate the model's accuracy select the INCORRECT classifications using the checkboxes in front of each post.\n
+    #         - Then, click on the 'Calculate Accuracy' button.\n
+    #         - Then, click on the 'Download data as CSV' to get the classifications and evaluation data as a .csv file.
+    #     """)
+    #     with gr.Row():
+    #         with gr.Column(scale=4):
+    #             file_input = gr.File(label="Upload CSV or TSV File", file_types=['.tsv', '.csv'])
+                
+    #         with gr.Column(scale=6):
+    #             text_field = gr.Textbox(label="Text field name", value="tweet_text")
+    #             event_model = gr.Dropdown(event_models, value=event_models[0], label="Select classification model")
+    #             ETCE_predict_button = gr.Button("Start Prediction")
+    #     with gr.Accordion("Prediction threshold", open=False):        
+    #         threshold = gr.Slider(0, 1, value=0, step=0.01, label="Prediction threshold", show_label=False, 
+    #                           info="This value sets a threshold by which texts classified flood or fire are accepted, \
+    #                               higher values makes the classifier stricter (CAUTION: A value of 1 will set all predictions as none)", interactive=True)
+        
+    #     with gr.Row(): 
+    #         with gr.Column():
+    #             gr.Markdown("""### Flood-related""")
+    #             flood_checkbox_output = gr.CheckboxGroup(label="Select ONLY incorrect classifications", interactive=True)
                 
+    #         with gr.Column():
+    #             gr.Markdown("""### Fire-related""")
+    #             fire_checkbox_output = gr.CheckboxGroup(label="Select ONLY incorrect classifications", interactive=True)
+                
+    #         with gr.Column():
+    #             gr.Markdown("""### None""")
+    #             none_checkbox_output = gr.CheckboxGroup(label="Select ONLY incorrect classifications", interactive=True) 
+
+    #     with gr.Row():
+    #         with gr.Column(scale=5):
+    #             gr.Markdown(r"""
+    #             Accuracy: is the model's ability to make correct predicitons.
+    #             It is the fraction of correct prediction out of the total predictions.
+                
+    #             $$
+    #             \text{Accuracy} = \frac{\text{Correct predictions}}{\text{All predictions}} * 100
+    #             $$
+                
+    #             Model Confidence: is the mean probabilty of each case 
+    #             belonging to their assigned classes. A value of 1 is best.
+    #             """, latex_delimiters=[{ "left": "$$", "right": "$$", "display": True }])
+    #             gr.Markdown("\n\n\n")
+    #             model_confidence = gr.Number(label="Model Confidence")
+                
+    #         with gr.Column(scale=5):
+    #             correct = gr.Number(label="Number of correct classifications")
+    #             incorrect = gr.Number(label="Number of incorrect classifications")
+    #             accuracy = gr.Number(label="Model Accuracy (%)")
+
+    #     ETCE_accuracy_button = gr.Button("Calculate Accuracy")
+    #     download_csv = gr.DownloadButton(visible=False)
+    #     num_posts = gr.Number(visible=False)
+    #     data = gr.DataFrame(visible=False) 
+    #     data_eval = gr.DataFrame(visible=False)
+
+        
 
 
          
@@ -360,6 +357,7 @@ with gr.Blocks(fill_width=True) as demo:
             }
         
         """
+    
     # Test event listeners
     T_predict_button.click(
         load_and_classify_csv_dataframe, 
@@ -382,22 +380,22 @@ with gr.Blocks(fill_width=True) as demo:
 
 
     # Button clicks ETC Eval
-    ETCE_predict_button.click(
-        load_and_classify_csv, 
-        inputs=[file_input, text_field, event_model, threshold], 
-        outputs=[flood_checkbox_output, fire_checkbox_output, none_checkbox_output, model_confidence, num_posts, data, QA_addqry_button, QA_run_button])
+    # ETCE_predict_button.click(
+    #     load_and_classify_csv, 
+    #     inputs=[file_input, text_field, event_model, threshold], 
+    #     outputs=[flood_checkbox_output, fire_checkbox_output, none_checkbox_output, model_confidence, num_posts, data, QA_addqry_button, QA_run_button])
     
-    ETCE_accuracy_button.click(
-        calculate_accuracy, 
-        inputs=[flood_checkbox_output, fire_checkbox_output, none_checkbox_output, num_posts, text_field, data], 
-        outputs=[incorrect, correct, accuracy, data_eval, download_csv])
+    # ETCE_accuracy_button.click(
+    #     calculate_accuracy, 
+    #     inputs=[flood_checkbox_output, fire_checkbox_output, none_checkbox_output, num_posts, text_field, data], 
+    #     outputs=[incorrect, correct, accuracy, data_eval, download_csv])
     
     
     # Button clicks QA
     QA_addqry_button.click(add_query, inputs=[query_inp, queries_state], outputs=[selected_queries, queries_state])
 
     QA_run_button.click(qa_summarise, 
-                    inputs=[selected_queries, qa_llm_model, text_field, data], ## XXX fix text_field 
+                    inputs=[selected_queries, qa_llm_model, T_text_field, T_data_ss_state],
                     outputs=[hsummary, qa_df])