Update app.py

app.py

@@ -21,30 +21,6 @@ def extract_table_from_markdown(markdown_text, table_start):
             table_content.append(line)
     return '\n'.join(table_content)
 
-# def markdown_table_to_df(table_content):
-#     """Convert markdown table to pandas DataFrame."""
-#     # Split the table content into lines
-#     lines = table_content.split('\n')
-    
-#     # Extract headers
-#     headers = [h.strip() for h in lines[0].split('|') if h.strip()]
-    
-#     # Extract data
-#     data = []
-#     for line in lines[2:]:  # Skip the header separator line
-#         row = [cell.strip() for cell in line.split('|') if cell.strip()]
-#         if row and len(row) == len(headers):  # Ensure row has the correct number of columns
-#             data.append(row)
-    
-#     # Create DataFrame
-#     df = pd.DataFrame(data, columns=headers)
-    
-#     # Convert numeric columns to float
-#     for col in df.columns:
-#         if col not in ["Model Name", "Publisher", "Open?", "Basemodel", "Matryoshka"]:
-#             df[col] = pd.to_numeric(df[col], errors='coerce')
-    
-#     return df
 
 def markdown_table_to_df(table_content):
     """Convert markdown table to pandas DataFrame."""
@@ -79,8 +55,29 @@ def setup_page():
     st.title("⚡ Swahili Text Embeddings Leaderboard (STEL)")
     st.image("https://raw.githubusercontent.com/username/repo/main/files/STEL.jpg", width=300)
 
+# def display_leaderboard(df):
+#     """Display the leaderboard."""
+#     st.header("📊 Leaderboard")
+    
+#     # Determine which non-benchmark columns are present
+#     present_non_benchmark_cols = [col for col in POSSIBLE_NON_BENCHMARK_COLS if col in df.columns]
+    
+#     # Add filters
+#     columns_to_filter = [col for col in df.columns if col not in present_non_benchmark_cols]
+#     selected_columns = st.multiselect("Select benchmarks to display:", columns_to_filter, default=columns_to_filter)
+    
+#     # Filter dataframe
+#     df_display = df[present_non_benchmark_cols + selected_columns]
+    
+#     # Display dataframe
+#     st.dataframe(df_display.style.format("{:.4f}", subset=[col for col in df_display.columns if df_display[col].dtype == 'float64']))
+    
+#     # Download buttons
+#     csv = df_display.to_csv(index=False)
+#     st.download_button(label="Download as CSV", data=csv, file_name="leaderboard.csv", mime="text/csv")
+
 def display_leaderboard(df):
-    """Display the leaderboard."""
+    """Display the leaderboard with compact columns."""
     st.header("📊 Leaderboard")
     
     # Determine which non-benchmark columns are present
@@ -93,13 +90,56 @@ def display_leaderboard(df):
     # Filter dataframe
     df_display = df[present_non_benchmark_cols + selected_columns]
     
-    # Display dataframe
-    st.dataframe(df_display.style.format("{:.4f}", subset=[col for col in df_display.columns if df_display[col].dtype == 'float64']))
+    # Define column widths
+    col_widths = {
+        "Model Name": 200,
+        "Publisher": 100,
+        "Open?": 60,
+        "Basemodel": 150,
+        "Matryoshka": 80,
+        "Dimension": 80,
+        "Average": 80
+    }
+    
+    # Set default width for benchmark columns
+    default_width = 80
+    
+    # Custom CSS to make the table more compact
+    st.markdown("""
+    <style>
+    .streamlit-expanderHeader {
+        font-size: 1em;
+    }
+    .stDataFrame {
+        font-size: 0.8em;
+    }
+    .stDataFrame td, .stDataFrame th {
+        white-space: nowrap;
+        overflow: hidden;
+        text-overflow: ellipsis;
+        max-width: 200px;
+    }
+    </style>
+    """, unsafe_allow_html=True)
+    
+    # Format the dataframe
+    df_styled = df_display.style.format({col: "{:.2f}" for col in df_display.columns if df_display[col].dtype in ['float64', 'int64']})
+    
+    # Set column widths
+    df_styled.set_properties(**{
+        'width': lambda x: f"{col_widths.get(x.name, default_width)}px",
+        'max-width': lambda x: f"{col_widths.get(x.name, default_width)}px",
+    })
+    
+    # Display the dataframe
+    st.dataframe(df_styled, height=400)
     
     # Download buttons
     csv = df_display.to_csv(index=False)
     st.download_button(label="Download as CSV", data=csv, file_name="leaderboard.csv", mime="text/csv")
 
+
+
 def display_evaluation():
     """Display the evaluation section."""
     st.header("🧪 Evaluation")