Update app.py

app.py

@@ -256,20 +256,20 @@ with demo:
 
         # 1. Benchmark Geral
         with gr.TabItem("📊 Benchmark Geral", id=tab_index):
-            # Colunas a exibir: T, Modelo, Média Geral, PLUE (remover outras médias)\
+            # Colunas a exibir: T, Modelo, Média Geral, PLUE
             general_cols_to_display = [
                 AutoEvalColumn.model_type_symbol.name, # T
                 AutoEvalColumn.model.name, # Modelo
                 AutoEvalColumn.average.name, # Média Geral
                 AutoEvalColumn.plue_avg.name, # Média PLUE
-                # Remover médias individuais das áreas NÃO PLUE (agora todas estão em PLUE)
-                # AutoEvalColumn.discurso_odio_avg.name,\
-                # AutoEvalColumn.economia_contabilidade_avg.name,\
-                # AutoEvalColumn.semantica_inferencia_avg.name,\
+                # Adicionar médias Energy/Reasoning se desejado (opcional)
+                # AutoEvalColumn.energy_avg.name, 
+                # AutoEvalColumn.reasoning_avg.name,
             ] 
             general_cols_to_display = [col for col in general_cols_to_display if col in LEADERBOARD_DF.columns]
             
-            # Colunas a ocultar: Todas as tasks + TODAS as médias de área individuais + detalhes
+            # Colunas a ocultar: Tasks + Médias de Área Individuais (PLUE e não-PLUE) + detalhes
+            # (Garantir que Energy/Reasoning avg estão aqui se não forem exibidas)
             general_hidden_cols = [task.name for task in Tasks] + \
                                   list(AREA_AVG_COLUMN_MAP.values()) + \
                                   [
@@ -291,69 +291,34 @@ with demo:
             )
         tab_index += 1
 
-        # 2. PLUE
+        # 2. PLUE (Agora apenas com as áreas originais + 3 adicionadas)
         with gr.TabItem("📚 PLUE", id=tab_index) as plue_tab:
-            # --- Lógica interna da aba PLUE (atualizada) ---
+            # --- Lógica interna da aba PLUE (ajustada) ---
             gr.Markdown("## Selecione a visualização PLUE:") 
-            # ATUALIZAR choices com base na nova PLUE_GROUP_AREAS
-            all_plue_options = [PLUE_GENERAL_VIEW_NAME] + sorted(PLUE_GROUP_AREAS) # Ordenar para consistência
+            # RECALCULAR choices e options com base na PLUE_GROUP_AREAS atualizada (sem Energy/Reasoning)
+            all_plue_options = [PLUE_GENERAL_VIEW_NAME] + sorted(PLUE_GROUP_AREAS) 
             plue_dropdown = gr.Dropdown(
                 choices=all_plue_options, 
                 label="Visualização PLUE", 
                 value=PLUE_GENERAL_VIEW_NAME 
             )
             
-            # Função auxiliar (ATUALIZAR lógica para PLUE_GENERAL_VIEW_NAME)
+            # Função auxiliar (lógica interna não muda, mas opera sobre PLUE_GROUP_AREAS atualizada)
             def get_plue_leaderboard_config(selected_option):
                 if selected_option == PLUE_GENERAL_VIEW_NAME:
-                    # Visão geral PLUE agora mostra TODAS as médias das áreas PLUE
-                    displayed_cols = [
-                        AutoEvalColumn.model_type_symbol.name,
-                        AutoEvalColumn.model.name,
-                    ] + [AREA_AVG_COLUMN_MAP[area] for area in PLUE_GROUP_AREAS if area in AREA_AVG_COLUMN_MAP]
-                    hidden_cols = [task.name for task in Tasks] + \
-                                  [AutoEvalColumn.average.name] + \
-                                  [
-                                      AutoEvalColumn.plue_avg.name, # Ocultar média PLUE agregada aqui
-                                      AutoEvalColumn.model_type.name,
-                                      AutoEvalColumn.architecture.name,
-                                      AutoEvalColumn.weight_type.name,
-                                      AutoEvalColumn.precision.name,
-                                      AutoEvalColumn.license.name,
-                                      AutoEvalColumn.params.name,
-                                      AutoEvalColumn.likes.name,
-                                      AutoEvalColumn.still_on_hub.name,
-                                      AutoEvalColumn.revision.name
-                                  ]
+                    displayed_cols = [AutoEvalColumn.model_type_symbol.name, AutoEvalColumn.model.name,] + [AREA_AVG_COLUMN_MAP[area] for area in PLUE_GROUP_AREAS if area in AREA_AVG_COLUMN_MAP]
+                    hidden_cols = [task.name for task in Tasks] + [avg_col for area, avg_col in AREA_AVG_COLUMN_MAP.items() if area not in PLUE_GROUP_AREAS] + [AutoEvalColumn.average.name] + [AutoEvalColumn.plue_avg.name, AutoEvalColumn.model_type.name, AutoEvalColumn.architecture.name, AutoEvalColumn.weight_type.name, AutoEvalColumn.precision.name, AutoEvalColumn.license.name, AutoEvalColumn.params.name, AutoEvalColumn.likes.name, AutoEvalColumn.still_on_hub.name, AutoEvalColumn.revision.name]
                     title = PLUE_GENERAL_VIEW_NAME
                 else:
-                    # Lógica para área específica (inalterada)
                     selected_area = selected_option
                     tasks_in_area = AREA_DEFINITIONS[selected_area]
-                    displayed_cols = [
-                        AutoEvalColumn.model_type_symbol.name,
-                        AutoEvalColumn.model.name,
-                    ] + [task.name for task in tasks_in_area]
-                    hidden_cols = list(AREA_AVG_COLUMN_MAP.values()) + \
-                                  [task.name for task in Tasks if task not in tasks_in_area] + \
-                                  [
-                                      AutoEvalColumn.plue_avg.name, # Ocultar média PLUE agregada aqui
-                                      AutoEvalColumn.average.name, # Ocultar média geral aqui também
-                                      AutoEvalColumn.model_type.name,
-                                      AutoEvalColumn.architecture.name,
-                                      AutoEvalColumn.weight_type.name,
-                                      AutoEvalColumn.precision.name,
-                                      AutoEvalColumn.license.name,
-                                      AutoEvalColumn.params.name,
-                                      AutoEvalColumn.likes.name,
-                                      AutoEvalColumn.still_on_hub.name,
-                                      AutoEvalColumn.revision.name
-                                  ] 
+                    displayed_cols = [AutoEvalColumn.model_type_symbol.name, AutoEvalColumn.model.name,] + [task.name for task in tasks_in_area]
+                    hidden_cols = list(AREA_AVG_COLUMN_MAP.values()) + [task.name for task in Tasks if task not in tasks_in_area] + [AutoEvalColumn.plue_avg.name, AutoEvalColumn.average.name, AutoEvalColumn.model_type.name, AutoEvalColumn.architecture.name, AutoEvalColumn.weight_type.name, AutoEvalColumn.precision.name, AutoEvalColumn.license.name, AutoEvalColumn.params.name, AutoEvalColumn.likes.name, AutoEvalColumn.still_on_hub.name, AutoEvalColumn.revision.name] 
                     title = selected_area
                 final_hidden_cols = [col for col in hidden_cols if col in LEADERBOARD_DF.columns]
                 return displayed_cols, final_hidden_cols, title
 
-            # Pré-renderização (ATUALIZAR loop com novas all_plue_options)
+            # Pré-renderização (ATUALIZAR loop e containers com novas all_plue_options)
             plue_containers = {} 
             for option in all_plue_options:
                 displayed_cols, hidden_cols, title = get_plue_leaderboard_config(option)
@@ -361,7 +326,7 @@ with demo:
                 with gr.Group(visible=is_visible) as plue_containers[option]:
                      create_leaderboard_component(LEADERBOARD_DF, displayed_cols=displayed_cols, hidden_cols=hidden_cols, title=title)
             
-            # Função de callback (inalterada, mas opera sobre novas opções)
+            # Função de callback (ATUALIZAR loop com novas all_plue_options)
             def switch_plue_view(selected_option):
                 update_list = []
                 for option in all_plue_options:
@@ -373,8 +338,32 @@ with demo:
             # --- Fim Lógica PLUE ---
         tab_index += 1
 
+        # 3. Energy 
+        with gr.TabItem("⚡️ Energy", id=tab_index):
+             # Exibir leaderboard com dados de Energy
+             energy_tasks = AREA_DEFINITIONS.get("Energy", [])
+             energy_cols = [AutoEvalColumn.model_type_symbol.name, AutoEvalColumn.model.name] + [t.name for t in energy_tasks]
+             energy_hidden = [t.name for t in Tasks if t not in energy_tasks] + \
+                             list(AREA_AVG_COLUMN_MAP.values()) + \
+                             [AutoEvalColumn.plue_avg.name, AutoEvalColumn.average.name] + \
+                             [c.name for c in fields(AutoEvalColumn) if c.name not in energy_cols and c.name != AutoEvalColumn.model_type_symbol.name and c.name != AutoEvalColumn.model.name ] # Detalhes
+             create_leaderboard_component(LEADERBOARD_DF, displayed_cols=energy_cols, hidden_cols=[c for c in energy_hidden if c in LEADERBOARD_DF.columns], title="Energy") 
+        tab_index += 1
+
+        # 4. Reasoning
+        with gr.TabItem("🤔 Reasoning", id=tab_index):
+             # Exibir leaderboard com dados de Reasoning
+             reasoning_tasks = AREA_DEFINITIONS.get("Reasoning", [])
+             reasoning_cols = [AutoEvalColumn.model_type_symbol.name, AutoEvalColumn.model.name] + [t.name for t in reasoning_tasks]
+             reasoning_hidden = [t.name for t in Tasks if t not in reasoning_tasks] + \
+                                list(AREA_AVG_COLUMN_MAP.values()) + \
+                                [AutoEvalColumn.plue_avg.name, AutoEvalColumn.average.name] + \
+                                [c.name for c in fields(AutoEvalColumn) if c.name not in reasoning_cols and c.name != AutoEvalColumn.model_type_symbol.name and c.name != AutoEvalColumn.model.name ] # Detalhes
+             create_leaderboard_component(LEADERBOARD_DF, displayed_cols=reasoning_cols, hidden_cols=[c for c in reasoning_hidden if c in LEADERBOARD_DF.columns], title="Reasoning") 
+        tab_index += 1
+
         # 5. Submit
-        with gr.TabItem("✉️✨ Submeta seu modelo aqui!", id=tab_index):
+        with gr.TabItem("📤 Submit aqui!", id=tab_index):
             with gr.Column():
                 with gr.Row():
                     gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text")