updates for dataset

app.py

@@ -9,10 +9,8 @@ from espn_ldbd.display.tabs import (
     create_leaderboard_tab,
     create_submit_tab
 )
-from espn_ldbd.leaderboard.data import (
-    LeaderboardData,
-    LeaderboardDataset
-)
+from espn_ldbd.leaderboard.data import LeaderboardData
+from espn_ldbd.leaderboard.dataset import LeaderboardDataset
 
 
 def create_app():

espn_ldbd/display/tabs.py

@@ -6,17 +6,11 @@ from espn_ldbd.leaderboard.data import LeaderboardData
 from espn_ldbd.leaderboard.dataset import LeaderboardDataset
 
 
-def create_leaderboard_tab(
-    task_name: str,
-    sub_tasks: dict,
-    leaderboard_data: LeaderboardData,
-    rows_per_page: int = 30
+def create_subtask_div(
+    
 ):
-    """Create a leaderboard tab with pagination controls"""
-
     # State to track current page
     page_state = gr.State(value=1)
-    selected_subtask = gr.State(value=0)
 
     with gr.Column():
         # Info section
@@ -46,7 +40,7 @@ def create_leaderboard_tab(
 
         # Dataframe display
         dataframe = gr.Dataframe(
-            value=leaderboard_data.get_paginated_data(task_name, 1, rows_per_page)[0],
+            value=leaderboard_data.get_paginated_data([task_name, sub_tasks], 1, rows_per_page)[0],
             interactive=False,
             wrap=True,
             column_widths=["5%", "25%", "15%", "15%", "15%", "15%", "10%"]
@@ -89,6 +83,10 @@ def create_leaderboard_tab(
         df, page_text, page = update_display(task_name, 1, rows)
         return df, page_text, page
 
+    def update_subtask(subtask_idx: int, rows):
+        df, page_text, page = update_display(task_name, 1, rows)
+        return df, page_text, page
+
     def change_rows_per_page(rows: int):
         """Change number of rows per page"""
         df, page_text, page = update_display(task_name, 1, rows)
@@ -113,6 +111,12 @@ def create_leaderboard_tab(
         outputs=[dataframe, page_info, page_state]
     )
 
+    subtask_dropdown.change(
+        fn=update_subtask,
+        inputs=[subtask_dropdown, rows_dropdown],
+        outputs=[dataframe, page_info, page_state]
+    )
+
     rows_dropdown.change(
         fn=change_rows_per_page,
         inputs=[rows_dropdown],
@@ -122,6 +126,54 @@ def create_leaderboard_tab(
     return dataframe, page_info, page_state
 
 
+def create_leaderboard_tab(
+    task_name: str,
+    sub_tasks: dict,
+    leaderboard_data: LeaderboardData,
+    rows_per_page: int = 30
+):
+    """Create a leaderboard tab with pagination controls"""
+
+    with gr.Column():
+        # Info section
+        gr.Markdown(f"## {task_name} Leaderboard")
+        gr.Markdown(f"Showing top performing models for {task_name.lower()} task")
+
+        with gr.Row():
+            with gr.Column():
+                subtask_dropdown = gr.Dropdown(
+                    label="Sub Tasks",
+                    choices=[(sub_tasks[x]["sub_task_title"], x) for x in range(len(sub_tasks))],
+                    value=0
+                )
+            with gr.Column():
+                pass
+
+            paginated_df, page_text, current_page = create_subtask_div(
+                
+            )
+    # def update_display(task: str, page: int, rows: int):
+    #     """Update the dataframe and page info"""
+    #     paginated_df, current_page, total_pages = leaderboard_data.get_paginated_data(
+    #         task, page, rows
+    #     )
+    #     page_text = f"Page {current_page} of {total_pages} ({len(leaderboard_data.get_data(task))} total entries)"
+    #     return paginated_df, page_text, current_page
+
+    def update_subtask(subtask_idx: int):
+        # df, page_text, page = update_display(task_name, 1, rows)
+        print(subtask_idx)
+        return  # df, page_text, page
+
+    subtask_dropdown.change(
+        fn=update_subtask,
+        inputs=[subtask_dropdown],
+        # outputs=[dataframe, page_info, page_state]
+    )
+
+    return   # dataframe, page_info, page_state
+
+
 def create_submit_tab(leaderboard_ds: LeaderboardDataset):
 
     def _submit_model(model_text):
@@ -159,7 +211,6 @@ def create_submit_tab(leaderboard_ds: LeaderboardDataset):
             f"{message}."
         )
         return model_text
-        
 
     with gr.Column():
         gr.Markdown("## Request a model here!")

espn_ldbd/leaderboard/data.py

@@ -1,3 +1,4 @@
+import os
 import math
 from typing import Dict, Tuple
 
@@ -8,6 +9,9 @@ class LeaderboardData:
     """Class to manage leaderboard data from datasets"""
     
     def __init__(self):
+        repo_id = os.environ.get("ESPNET_DB", None)
+        assert repo_id is not None
+        self._repo_id = repo_id
         self.data_cache: Dict[str, pd.DataFrame] = {}
         self.current_page: Dict[str, int] = {}
         
@@ -90,10 +94,10 @@ class LeaderboardData:
     def get_data(self, task_name: str, force_reload: bool = False) -> pd.DataFrame:
         """Get data for a specific task, with caching"""
         if task_name not in self.data_cache or force_reload:
-            self.data_cache[task_name] = self.load_data_from_dataset("espnet/leaderboard", task_name)
+            self.data_cache[task_name] = self.load_data_from_dataset(self._repo_id, task_name)
             self.current_page[task_name] = 1
         return self.data_cache[task_name]
-    
+
     def get_paginated_data(self, task_name: str, page: int, rows_per_page: int = 30) -> Tuple[pd.DataFrame, int, int]:
         """Get paginated data for display"""
         df = self.get_data(task_name)