check

app.py

@@ -1,5 +1,5 @@
 import gradio as gr
-from gradio_leaderboard import Leaderboard, SelectColumns
+from gradio_leaderboard import Leaderboard, ColumnFilter, SelectColumns
 import pandas as pd
 from apscheduler.schedulers.background import BackgroundScheduler
 from huggingface_hub import snapshot_download
@@ -74,43 +74,49 @@ LEADERBOARD_DF_CDM_FI = get_leaderboard_df(EVAL_RESULTS_PATH_CDM_FI, COLS, BENCH
 def init_leaderboard(dataframe):
     if dataframe is None or dataframe.empty:
         print("Warning: Empty dataframe provided to leaderboard")
-        # Return a simple empty dataframe instead of trying to create a custom one
-        return gr.Dataframe(value=pd.DataFrame())
+        return gr.Dataframe(
+            headers=COLS, datatype=[c.type for c in fields(AutoEvalColumn)], label="No results available"
+        )
 
     print(f"Initializing leaderboard with {len(dataframe)} rows")
     print(f"Columns: {dataframe.columns.tolist()}")
 
-    # Convert dataframe to ensure proper types
-    for col in dataframe.columns:
-        if col in ["average", "params"] + [t.value.col_name for t in Tasks]:
-            dataframe[col] = pd.to_numeric(dataframe[col], errors="coerce")
-        elif col == "still_on_hub":
-            dataframe[col] = dataframe[col].astype(bool)
-        else:
-            dataframe[col] = dataframe[col].astype(str)
-
     try:
-        return gr.Dataframe(value=dataframe, interactive=False, wrap=True)
+        return Leaderboard(
+            value=dataframe,
+            headers=COLS,  # Explicitly specify headers
+            datatype=[c.type for c in fields(AutoEvalColumn)],
+            select_columns=SelectColumns(
+                default_selection=[c.name for c in fields(AutoEvalColumn) if c.displayed_by_default],
+                cant_deselect=[c.name for c in fields(AutoEvalColumn) if c.never_hidden],
+                label="Select Columns to Display:",
+            ),
+            search_columns=[AutoEvalColumn.model.name],
+            interactive=False,
+        )
     except Exception as e:
         print(f"Error initializing leaderboard: {e}")
-        return gr.Dataframe(value=pd.DataFrame())
+        return gr.Dataframe(
+            value=dataframe,
+            headers=COLS,
+            datatype=[c.type for c in fields(AutoEvalColumn)],
+            label="Error loading interactive leaderboard",
+        )
 
 
-# Initialize the app
 demo = gr.Blocks(css=custom_css)
-
 with demo:
     gr.HTML(TITLE)
     gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
 
-    with gr.Tabs() as tabs:
-        with gr.Tab("MIMIC CDM"):
+    with gr.Tabs(elem_classes="tab-buttons") as tabs:
+        with gr.TabItem("MIMIC CDM", elem_id="llm-benchmark-tab-table", id=0):
             leaderboard_cdm = init_leaderboard(LEADERBOARD_DF_CDM)
 
-        with gr.Tab("MIMIC CDM FI"):
+        with gr.TabItem("MIMIC CDM FI", elem_id="llm-benchmark-tab-table", id=1):
             leaderboard_cdm_fi = init_leaderboard(LEADERBOARD_DF_CDM_FI)
 
-        with gr.Tab("📝 About"):
+        with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=2):
             gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
 
     with gr.Row():
@@ -123,10 +129,7 @@ with demo:
                 show_copy_button=True,
             )
 
-# Start the scheduler
 scheduler = BackgroundScheduler()
 scheduler.add_job(restart_space, "interval", seconds=1800)
 scheduler.start()
-
-# Launch the app
-demo.queue(default_concurrency_limit=40).launch()
+demo.queue(default_concurrency_limit=40).launch(share=True)

src/populate.py

@@ -19,28 +19,11 @@ def get_leaderboard_df(results_path: str, cols: list, benchmark_cols: list) -> p
 
     try:
         df = pd.DataFrame.from_records(all_data_json)
-
-        # Ensure all required columns exist with proper types
+        # Ensure all required columns exist
         for col in cols:
             if col not in df.columns:
                 df[col] = None
 
-        # Convert numeric columns
-        numeric_cols = ["average", "params"] + [t.value.col_name for t in Tasks]
-        for col in numeric_cols:
-            if col in df.columns:
-                df[col] = pd.to_numeric(df[col], errors="coerce")
-
-        # Convert boolean columns
-        if "still_on_hub" in df.columns:
-            df["still_on_hub"] = df["still_on_hub"].astype(bool)
-
-        # Convert string columns
-        string_cols = ["model", "architecture"]
-        for col in string_cols:
-            if col in df.columns:
-                df[col] = df[col].astype(str)
-
         df = df.sort_values(by=[AutoEvalColumn.average.name], ascending=False)
         df = df[cols].round(decimals=2)