debug

app.py

@@ -128,122 +128,122 @@ from src.about import TasksMib_Subgraph
 
 
 
-# def init_leaderboard_mib_subgraph(dataframe, track):
-#     """Initialize the subgraph leaderboard with grouped column selection by benchmark."""
-#     if dataframe is None or dataframe.empty:
-#         raise ValueError("Leaderboard DataFrame is empty or None.")
-
-#     print("\nDebugging DataFrame columns:", dataframe.columns.tolist())
-    
-#     # Create groups of columns by benchmark
-#     benchmark_groups = []
-    
-#     # For each benchmark in our TasksMib_Subgraph enum...
-#     for task in TasksMib_Subgraph:
-#         benchmark = task.value.benchmark
-#         # Get all valid columns for this benchmark's models
-#         benchmark_cols = [
-#             f"{benchmark}_{model}"
-#             for model in task.value.models
-#             if f"{benchmark}_{model}" in dataframe.columns
-#         ]
-#         if benchmark_cols:  # Only add if we have valid columns
-#             benchmark_groups.append(benchmark_cols)
-#             print(f"\nBenchmark group for {benchmark}:", benchmark_cols)
-
-#     # Create model groups as well
-#     model_groups = []
-#     all_models = list(set(model for task in TasksMib_Subgraph for model in task.value.models))
-    
-#     # For each unique model...
-#     for model in all_models:
-#         # Get all valid columns for this model across benchmarks
-#         model_cols = [
-#             f"{task.value.benchmark}_{model}"
-#             for task in TasksMib_Subgraph
-#             if model in task.value.models
-#             and f"{task.value.benchmark}_{model}" in dataframe.columns
-#         ]
-#         if model_cols:  # Only add if we have valid columns
-#             model_groups.append(model_cols)
-#             print(f"\nModel group for {model}:", model_cols)
-
-#     # Combine all groups
-#     all_groups = benchmark_groups + model_groups
-    
-#     # Flatten groups for default selection (show everything initially)
-#     all_columns = [col for group in all_groups for col in group]
-#     print("\nAll available columns:", all_columns)
-
-#     return Leaderboard(
-#         value=dataframe,
-#         datatype=[c.type for c in fields(AutoEvalColumn_mib_subgraph)],
-#         select_columns=SelectColumns(
-#             default_selection=all_columns,  # Show all columns initially
-#             label="Select Results:"
-#         ),
-#         search_columns=["Method"],
-#         hide_columns=[],
-#         interactive=False,
-#     )
-
 def init_leaderboard_mib_subgraph(dataframe, track):
-    """Initialize the subgraph leaderboard with group-based column selection."""
+    """Initialize the subgraph leaderboard with grouped column selection by benchmark."""
     if dataframe is None or dataframe.empty:
         raise ValueError("Leaderboard DataFrame is empty or None.")
 
     print("\nDebugging DataFrame columns:", dataframe.columns.tolist())
     
-    # Create selection mapping for benchmark groups
-    selection_mapping = {}
+    # Create groups of columns by benchmark
+    benchmark_groups = []
     
-    # Create benchmark groups with descriptive names
+    # For each benchmark in our TasksMib_Subgraph enum...
     for task in TasksMib_Subgraph:
         benchmark = task.value.benchmark
-        # Get all columns for this benchmark's models
+        # Get all valid columns for this benchmark's models
         benchmark_cols = [
             f"{benchmark}_{model}"
             for model in task.value.models
             if f"{benchmark}_{model}" in dataframe.columns
         ]
-        if benchmark_cols:
-            # Use a descriptive group name as the key
-            group_name = f"Benchmark: {benchmark.upper()}"
-            selection_mapping[group_name] = benchmark_cols
-            print(f"\n{group_name} maps to:", benchmark_cols)
+        if benchmark_cols:  # Only add if we have valid columns
+            benchmark_groups.append(benchmark_cols)
+            print(f"\nBenchmark group for {benchmark}:", benchmark_cols)
 
-    # Create model groups with descriptive names
+    # Create model groups as well
+    model_groups = []
     all_models = list(set(model for task in TasksMib_Subgraph for model in task.value.models))
+    
+    # For each unique model...
     for model in all_models:
-        # Get all columns for this model across benchmarks
+        # Get all valid columns for this model across benchmarks
         model_cols = [
             f"{task.value.benchmark}_{model}"
             for task in TasksMib_Subgraph
             if model in task.value.models
             and f"{task.value.benchmark}_{model}" in dataframe.columns
         ]
-        if model_cols:
-            # Use a descriptive group name as the key
-            group_name = f"Model: {model}"
-            selection_mapping[group_name] = model_cols
-            print(f"\n{group_name} maps to:", model_cols)
+        if model_cols:  # Only add if we have valid columns
+            model_groups.append(model_cols)
+            print(f"\nModel group for {model}:", model_cols)
 
-    # The selection options are the group names
-    selection_options = list(selection_mapping.keys())
-    print("\nSelection options:", selection_options)
+    # Combine all groups
+    all_groups = benchmark_groups + model_groups
+    
+    # Flatten groups for default selection (show everything initially)
+    all_columns = [col for group in all_groups for col in group]
+    print("\nAll available columns:", all_columns)
 
     return Leaderboard(
         value=dataframe,
         datatype=[c.type for c in fields(AutoEvalColumn_mib_subgraph)],
         select_columns=SelectColumns(
-            default_selection=selection_options,  # Show all groups by default
-            label="Select Benchmark or Model Groups:"
+            default_selection=all_columns,  # Show all columns initially
+            label="Select Results:"
         ),
         search_columns=["Method"],
         hide_columns=[],
         interactive=False,
     )
 
+# def init_leaderboard_mib_subgraph(dataframe, track):
+#     """Initialize the subgraph leaderboard with group-based column selection."""
+#     if dataframe is None or dataframe.empty:
+#         raise ValueError("Leaderboard DataFrame is empty or None.")
+
+#     print("\nDebugging DataFrame columns:", dataframe.columns.tolist())
+    
+#     # Create selection mapping for benchmark groups
+#     selection_mapping = {}
+    
+#     # Create benchmark groups with descriptive names
+#     for task in TasksMib_Subgraph:
+#         benchmark = task.value.benchmark
+#         # Get all columns for this benchmark's models
+#         benchmark_cols = [
+#             f"{benchmark}_{model}"
+#             for model in task.value.models
+#             if f"{benchmark}_{model}" in dataframe.columns
+#         ]
+#         if benchmark_cols:
+#             # Use a descriptive group name as the key
+#             group_name = f"Benchmark: {benchmark.upper()}"
+#             selection_mapping[group_name] = benchmark_cols
+#             print(f"\n{group_name} maps to:", benchmark_cols)
+
+#     # Create model groups with descriptive names
+#     all_models = list(set(model for task in TasksMib_Subgraph for model in task.value.models))
+#     for model in all_models:
+#         # Get all columns for this model across benchmarks
+#         model_cols = [
+#             f"{task.value.benchmark}_{model}"
+#             for task in TasksMib_Subgraph
+#             if model in task.value.models
+#             and f"{task.value.benchmark}_{model}" in dataframe.columns
+#         ]
+#         if model_cols:
+#             # Use a descriptive group name as the key
+#             group_name = f"Model: {model}"
+#             selection_mapping[group_name] = model_cols
+#             print(f"\n{group_name} maps to:", model_cols)
+
+#     # The selection options are the group names
+#     selection_options = list(selection_mapping.keys())
+#     print("\nSelection options:", selection_options)
+
+#     return Leaderboard(
+#         value=dataframe,
+#         datatype=[c.type for c in fields(AutoEvalColumn_mib_subgraph)],
+#         select_columns=SelectColumns(
+#             default_selection=selection_options,  # Show all groups by default
+#             label="Select Benchmark or Model Groups:"
+#         ),
+#         search_columns=["Method"],
+#         hide_columns=[],
+#         interactive=False,
+#     )
+