fix csv loading

app.py

@@ -232,55 +232,33 @@ multilingual_df = create_multilingual_dataframe()
 def create_longform_dataframe():
     """Create longform dataframe from CSV data"""
     if longform_csv_path is None or not longform_csv_path.exists():
-        print("Longform CSV not found, creating sample data")
-        # Fallback to sample data if CSV not available
-        longform_data = []
-        sample_models = [
-            {"model": "openai/whisper-large-v3", "earnings21": 8.2, "mustc": 12.4, "rtfx": 2.1},
-            {"model": "openai/whisper-large-v2", "earnings21": 9.1, "mustc": 13.8, "rtfx": 1.8},
-            {"model": "nvidia/canary-1b", "earnings21": 7.5, "mustc": 11.2, "rtfx": 3.2},
-            {"model": "microsoft/speecht5_asr", "earnings21": 15.3, "mustc": 18.7, "rtfx": 1.4},
-        ]
-        
-        for model_data in sample_models:
-            row = {
-                "Model": make_clickable_model(model_data["model"]),
-                "Average WER ⬇️": round((model_data["earnings21"] + model_data["mustc"]) / 2, 2),
-                "RTFx ⬆️️": model_data["rtfx"] if model_data["rtfx"] > 0 else "NA",
-                "Earnings21": model_data["earnings21"],
-                "MustC": model_data["mustc"]
-            }
-            longform_data.append(row)
+        raise Exception(f"Longform CSV file {longform_csv_path} does not exist locally")
+    
+    # Load from CSV
+    longform_raw_df = pd.read_csv(longform_csv_path)
+    longform_data = []
+    
+    for _, row_data in longform_raw_df.iterrows():
+        model_name = row_data['model']
         
-        longform_df = pd.DataFrame(longform_data)
-    else:
-        # Load from CSV
-        longform_raw_df = pd.read_csv(longform_csv_path)
-        longform_data = []
+        # Get values from CSV, similar to other tabs
+        earnings21_wer = row_data.get('earnings21_wer', -1)
+        mustc_wer = row_data.get('mustc_wer', -1)
+        rtfx_value = row_data.get('rtfx', 0)
         
-        for _, row_data in longform_raw_df.iterrows():
-            model_name = row_data['model']
-            
-            # Get values from CSV, similar to other tabs
-            earnings21_wer = row_data.get('earnings21_wer', -1)
-            mustc_wer = row_data.get('mustc_wer', -1)
-            rtfx_value = row_data.get('rtfx', 0)
-            
-            # Calculate average WER from available datasets
-            available_wers = [w for w in [earnings21_wer, mustc_wer] if w != -1 and w > 0]
-            avg_wer = round(np.mean(available_wers), 2) if available_wers else 0.0
-            
-            row = {
-                "Model": make_clickable_model(model_name),
-                "Average WER ⬇️": avg_wer,
-                "RTFx ⬆️️": rtfx_value if rtfx_value > 0 else "NA",
-                "Earnings21": earnings21_wer if earnings21_wer != -1 else "NA",
-                "MustC": mustc_wer if mustc_wer != -1 else "NA"
-            }
-            longform_data.append(row)
+        # Calculate average WER from available datasets
+        available_wers = [w for w in [earnings21_wer, mustc_wer] if w != -1 and w > 0]
+        avg_wer = round(np.mean(available_wers), 2) if available_wers else 0.0
         
-        longform_df = pd.DataFrame(longform_data)
+        row = {
+            "Model": make_clickable_model(model_name),
+            "Average WER ⬇️": avg_wer,
+            "RTFx ⬆️️": rtfx_value if rtfx_value > 0 else "NA",
+            "Earnings21": earnings21_wer if earnings21_wer != -1 else "NA",
+        }
+        longform_data.append(row)
     
+    longform_df = pd.DataFrame(longform_data)
     longform_df = longform_df.sort_values(by='Average WER ⬇️')
     return longform_df