Update app.py

make model name cliquable and around the metrics

app.py

@@ -28,13 +28,7 @@ else:
         leaderboard_df.to_csv(leaderboard_file, index=False)
 
 def normalize_text(text):
-    """
-    Normalize text for WER/CER calculation:
-    - Convert to lowercase
-    - Remove punctuation
-    - Replace multiple spaces with single space
-    - Strip leading/trailing spaces
-    """
+    """Normalize text for WER/CER calculation"""
     if not isinstance(text, str):
         text = str(text)
     
@@ -92,21 +86,42 @@ def calculate_metrics(predictions_df):
     avg_wer = sum(item["wer"] for item in results) / len(results)
     avg_cer = sum(item["cer"] for item in results) / len(results)
     
+    # Calculate weighted average metrics based on reference length
     weighted_wer = sum(item["wer"] * item["ref_word_count"] for item in results) / total_ref_words
     weighted_cer = sum(item["cer"] * item["ref_char_count"] for item in results) / total_ref_chars
     
     return avg_wer, avg_cer, weighted_wer, weighted_cer, results
 
-def add_ranking_numbers(df, sort_by="Combined_Score"):
-    """Add ranking numbers to the dataframe based on the sort column"""
+def format_as_percentage(value):
+    """Convert decimal to percentage with 2 decimal places"""
+    return f"{value * 100:.2f}%"
+
+def make_clickable_model(model_name):
+    """Format model name as clickable link to Hugging Face hub"""
+    link = f"https://huggingface.co/{model_name}"
+    return f'<a href="{link}" target="_blank" style="text-decoration: underline;">{model_name}</a>'
+
+def prepare_leaderboard_for_display(df, sort_by="Combined_Score"):
+    """Format leaderboard for display with ranking and percentages"""
     if len(df) == 0:
-        return pd.DataFrame(columns=["Rank"] + list(df.columns))
+        return pd.DataFrame(columns=["Rank", "Model_Name", "WER (%)", "CER (%)", "Combined_Score (%)", "timestamp"])
     
 
-    sorted_df = df.sort_values(sort_by)    
-    sorted_df.insert(0, "Rank", range(1, len(sorted_df) + 1))
+    display_df = df.copy()
+    
+    display_df = display_df.sort_values(sort_by)
+    
+    display_df.insert(0, "Rank", range(1, len(display_df) + 1))
     
-    return sorted_df
+    for col in ["WER", "CER", "Combined_Score"]:
+        if col in display_df.columns:
+            display_df[f"{col} (%)"] = display_df[col].apply(lambda x: f"{x * 100:.2f}")
+            display_df = display_df.drop(col, axis=1)
+    
+    if "Model_Name" in display_df.columns:
+        display_df["Model_Name"] = display_df["Model_Name"].apply(make_clickable_model)
+    
+    return display_df
 
 def update_ranking(method):
     """Update leaderboard ranking based on selected method"""
@@ -122,10 +137,10 @@ def update_ranking(method):
         elif method == "CER Only":
             sort_column = "CER"
         
-        return add_ranking_numbers(current_lb, sort_column)
+        return prepare_leaderboard_for_display(current_lb, sort_column)
         
     except Exception:
-        return pd.DataFrame(columns=["Rank", "Model_Name", "WER", "CER", "Combined_Score", "timestamp"])
+        return pd.DataFrame(columns=["Rank", "Model_Name", "WER (%)", "CER (%)", "Combined_Score (%)", "timestamp"])
 
 def process_submission(model_name, csv_file):
     try:
@@ -163,6 +178,7 @@ def process_submission(model_name, csv_file):
         leaderboard = pd.read_csv(leaderboard_file)
         timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
         
+        # Calculate combined score (70% WER, 30% CER)
         combined_score = avg_wer * 0.7 + avg_cer * 0.3
         
         new_entry = pd.DataFrame(
@@ -170,13 +186,13 @@ def process_submission(model_name, csv_file):
             columns=["Model_Name", "WER", "CER", "Combined_Score", "timestamp"]
         )
         
-       
+
         updated_leaderboard = pd.concat([leaderboard, new_entry]).sort_values("Combined_Score")
         updated_leaderboard.to_csv(leaderboard_file, index=False)
         
-        ranked_leaderboard = add_ranking_numbers(updated_leaderboard)
+        display_leaderboard = prepare_leaderboard_for_display(updated_leaderboard)
         
-        return f"Submission processed successfully! WER: {avg_wer:.4f}, CER: {avg_cer:.4f}, Combined Score: {combined_score:.4f}", ranked_leaderboard
+        return f"Submission processed successfully! WER: {format_as_percentage(avg_wer)}, CER: {format_as_percentage(avg_cer)}, Combined Score: {format_as_percentage(combined_score)}", display_leaderboard
         
     except Exception as e:
         return f"Error processing submission: {str(e)}", None
@@ -184,7 +200,7 @@ def process_submission(model_name, csv_file):
 with gr.Blocks(title="Bambara ASR Leaderboard") as demo:
     gr.Markdown(
         """
-        # Bambara ASR Leaderboard
+        # 🇲🇱 Bambara ASR Leaderboard
         
         This leaderboard ranks and evaluates speech recognition models for the Bambara language.
         Models are ranked based on a combined score of WER and CER metrics.
@@ -199,10 +215,9 @@ with gr.Blocks(title="Bambara ASR Leaderboard") as demo:
                 if "Combined_Score" not in current_leaderboard.columns:
                     current_leaderboard["Combined_Score"] = current_leaderboard["WER"] * 0.7 + current_leaderboard["CER"] * 0.3
                 
-                
-                current_leaderboard = add_ranking_numbers(current_leaderboard.sort_values("Combined_Score"))
+                display_leaderboard = prepare_leaderboard_for_display(current_leaderboard)
             except Exception:
-                current_leaderboard = pd.DataFrame(columns=["Rank", "Model_Name", "WER", "CER", "Combined_Score", "timestamp"])
+                display_leaderboard = pd.DataFrame(columns=["Rank", "Model_Name", "WER (%)", "CER (%)", "Combined_Score (%)", "timestamp"])
             
             gr.Markdown("### Current ASR Model Rankings")
             
@@ -213,7 +228,7 @@ with gr.Blocks(title="Bambara ASR Leaderboard") as demo:
             )
             
             leaderboard_view = gr.DataFrame(
-                value=current_leaderboard,
+                value=display_leaderboard,
                 interactive=False,
                 label="Models are ranked by selected metric - lower is better"
             )
@@ -227,9 +242,9 @@ with gr.Blocks(title="Bambara ASR Leaderboard") as demo:
             gr.Markdown(
                 """
                 ## Metrics Explanation
-                - **WER**: Word Error Rate (lower is better) - measures word-level accuracy
-                - **CER**: Character Error Rate (lower is better) - measures character-level accuracy
-                - **Combined Score**: Weighted average of WER (70%) and CER (30%) - provides a balanced evaluation
+                - **WER (%)**: Word Error Rate (lower is better) - measures word-level accuracy
+                - **CER (%)**: Character Error Rate (lower is better) - measures character-level accuracy
+                - **Combined Score (%)**: Weighted average of WER (70%) and CER (30%) - provides a balanced evaluation
                 """
             )
         
@@ -251,7 +266,7 @@ with gr.Blocks(title="Bambara ASR Leaderboard") as demo:
             output_msg = gr.Textbox(label="Status", interactive=False)
             leaderboard_display = gr.DataFrame(
                 label="Updated Leaderboard",
-                value=current_leaderboard,
+                value=display_leaderboard,
                 interactive=False
             )
             
@@ -262,4 +277,4 @@ with gr.Blocks(title="Bambara ASR Leaderboard") as demo:
             )
 
 if __name__ == "__main__":
-    demo.launch()  
+    demo.launch()