Update app.py

app.py

@@ -56,25 +56,26 @@ with block:
         # table jmmmu bench
         with gr.TabItem("🏅 PROBE Leaderboard", elem_id="probe-benchmark-tab-table", id=1):
 
-            with gr.Row(show_progress=True):
-                method_names = pd.read_csv(CSV_RESULT_PATH)['method_name'].unique().tolist()
-                metric_names = pd.read_csv(CSV_RESULT_PATH).columns.tolist()
-                metrics_with_method = metric_names.copy()
-                metric_names.remove('method_name')  # Remove method_name from the metric options
-            
-                # Leaderboard section with method and metric selectors
-                leaderboard_method_selector = gr.CheckboxGroup(
-                    choices=method_names, label="Select method_names for Leaderboard", value=method_names, interactive=True
-                )
-                leaderboard_metric_selector = gr.CheckboxGroup(
-                    choices=metric_names, label="Select Metrics for Leaderboard", value=metric_names, interactive=True
-                )
 
-                # Display the filtered leaderboard
-                baseline_value = get_baseline_df(method_names, metric_names)
-                baseline_header = ["method_name"] + metric_names
-                baseline_datatype = ['markdown'] + ['number'] * len(metric_names)
-        
+            method_names = pd.read_csv(CSV_RESULT_PATH)['method_name'].unique().tolist()
+            metric_names = pd.read_csv(CSV_RESULT_PATH).columns.tolist()
+            metrics_with_method = metric_names.copy()
+            metric_names.remove('method_name')  # Remove method_name from the metric options
+            
+            # Leaderboard section with method and metric selectors
+            leaderboard_method_selector = gr.CheckboxGroup(
+                choices=method_names, label="Select method_names for Leaderboard", value=method_names, interactive=True
+            )
+            leaderboard_metric_selector = gr.CheckboxGroup(
+                choices=metric_names, label="Select Metrics for Leaderboard", value=metric_names, interactive=True
+            )
+
+            # Display the filtered leaderboard
+            baseline_value = get_baseline_df(method_names, metric_names)
+            baseline_header = ["method_name"] + metric_names
+            baseline_datatype = ['markdown'] + ['number'] * len(metric_names)
+
+            with gr.Row(show_progress=True)
                 data_component = gr.components.Dataframe(
                     value=baseline_value,
                     headers=baseline_header,
@@ -84,51 +85,50 @@ with block:
                     visible=True,
                 )
     
-                # Update leaderboard when method/metric selection changes
-                leaderboard_method_selector.change(
-                    update_leaderboard, 
-                    inputs=[leaderboard_method_selector, leaderboard_metric_selector], 
-                    outputs=data_component
-                )
-                leaderboard_metric_selector.change(
-                    update_leaderboard, 
-                    inputs=[leaderboard_method_selector, leaderboard_metric_selector], 
-                    outputs=data_component
-                )
-
-            with gr.Row(variant='panel', show_progress=True):
-
-                # Dropdown for benchmark type
-                benchmark_type_selector = gr.Dropdown(choices=list(benchmark_specific_metrics.keys()), label="Select Benchmark Type")
-
-                # Dynamic selectors
-                x_metric_selector = gr.Dropdown(choices=[], label="Select X-axis Metric", visible=False)
-                y_metric_selector = gr.Dropdown(choices=[], label="Select Y-axis Metric", visible=False)
-                aspect_type_selector = gr.Dropdown(choices=[], label="Select Aspect Type", visible=False)
-                dataset_type_selector = gr.Dropdown(choices=[], label="Select Dataset Type", visible=False)
-                dataset_selector = gr.Dropdown(choices=[], label="Select Dataset", visible=False)
-                single_metric_selector = gr.Dropdown(choices=[], label="Select Metric", visible=False)
-
-                method_selector = gr.CheckboxGroup(choices=method_names, label="Select methods to visualize", interactive=True, value=method_names)
+            # Update leaderboard when method/metric selection changes
+            leaderboard_method_selector.change(
+                update_leaderboard, 
+                inputs=[leaderboard_method_selector, leaderboard_metric_selector], 
+                outputs=data_component
+            )
+            leaderboard_metric_selector.change(
+                update_leaderboard, 
+                inputs=[leaderboard_method_selector, leaderboard_metric_selector], 
+                outputs=data_component
+            )
+
+
+            # Dropdown for benchmark type
+            benchmark_type_selector = gr.Dropdown(choices=list(benchmark_specific_metrics.keys()), label="Select Benchmark Type")
+
+            # Dynamic selectors
+            x_metric_selector = gr.Dropdown(choices=[], label="Select X-axis Metric", visible=False)
+            y_metric_selector = gr.Dropdown(choices=[], label="Select Y-axis Metric", visible=False)
+            aspect_type_selector = gr.Dropdown(choices=[], label="Select Aspect Type", visible=False)
+            dataset_type_selector = gr.Dropdown(choices=[], label="Select Dataset Type", visible=False)
+            dataset_selector = gr.Dropdown(choices=[], label="Select Dataset", visible=False)
+            single_metric_selector = gr.Dropdown(choices=[], label="Select Metric", visible=False)
+
+            method_selector = gr.CheckboxGroup(choices=method_names, label="Select methods to visualize", interactive=True, value=method_names)
                 
-                # Button to draw the plot for the selected benchmark
+            # Button to draw the plot for the selected benchmark
 
-                plot_button = gr.Button("Plot")
+            plot_button = gr.Button("Plot")
                 
-                plot_output = gr.Image(label="Plot")
+            plot_output = gr.Image(label="Plot")
                 
-                # Update selectors when benchmark type changes
-                benchmark_type_selector.change(
-                    update_metric_choices,
-                    inputs=[benchmark_type_selector],
-                    outputs=[x_metric_selector, y_metric_selector, aspect_type_selector, dataset_type_selector, dataset_selector, single_metric_selector]
-                )
+            # Update selectors when benchmark type changes
+            benchmark_type_selector.change(
+                update_metric_choices,
+                inputs=[benchmark_type_selector],
+                outputs=[x_metric_selector, y_metric_selector, aspect_type_selector, dataset_type_selector, dataset_selector, single_metric_selector]
+            )
                 
-                plot_button.click(
-                    benchmark_plot,
-                    inputs=[benchmark_type_selector, method_selector, x_metric_selector, y_metric_selector, aspect_type_selector, dataset_type_selector, dataset_selector, single_metric_selector],
-                    outputs=plot_output
-                )
+            plot_button.click(
+                benchmark_plot,
+                inputs=[benchmark_type_selector, method_selector, x_metric_selector, y_metric_selector, aspect_type_selector, dataset_type_selector, dataset_selector, single_metric_selector],
+                outputs=plot_output
+            )
             
         with gr.TabItem("📝 About", elem_id="probe-benchmark-tab-table", id=2):
             with gr.Row():