Update space1

src/populate.py

@@ -34,27 +34,36 @@ def get_leaderboard_df(results_path: str, requests_path: str, cols: list, benchm
                 "Title search rate (%)": 0
             }])], ignore_index=True)
 
+    def extract_first(value):
+        if isinstance(value, (list, np.ndarray)):
+            return value[0] if len(value) > 0 else 0
+        elif isinstance(value, (int, float)):
+            return value
+        else:
+            return 0
+
+    df["(T1) Precision (%)"] = df["(T1) Precision (%)"].apply(extract_first)
     # 将数组转标量，空数组变为 0
-    df["(T1) Precision (%)"] = df["(T1) Precision (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T1) Title_search_rate (%)"] = df["(T1) Title_search_rate (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T1) Overlap (%)"] = df["(T1) Overlap (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T1) Precision (First Author) (%)"] = df["(T1) Precision (First Author) (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T1) Overlap (First Author) (%)"] = df["(T1) Overlap (First Author) (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
+    df["(T1) Precision (%)"] = df["(T1) Precision (%)"].apply(extract_first)
+    df["(T1) Title_search_rate (%)"] = df["(T1) Title_search_rate (%)"].apply(extract_first)
+    df["(T1) Overlap (%)"] = df["(T1) Overlap (%)"].apply(extract_first)
+    df["(T1) Precision (First Author) (%)"] = df["(T1) Precision (First Author) (%)"].apply(extract_first)
+    df["(T1) Overlap (First Author) (%)"] = df["(T1) Overlap (First Author) (%)"].apply(extract_first)
     # Task 2
-    df["(T2) Similarity (%)"] = df["(T2) Similarity (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T2) Entailment (TRUE %)"] = df["(T2) Entailment (TRUE %)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T2) Entailment (GPT-4o %)"] = df["(T2) Entailment (GPT-4o %)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T2) ROUGE-1 (%)"] = df["(T2) ROUGE-1 (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T2) ROUGE-2 (%)"] = df["(T2) ROUGE-2 (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T2) ROUGE-L (%)"] = df["(T2) ROUGE-L (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
+    df["(T2) Similarity (%)"] = df["(T2) Similarity (%)"].apply(extract_first)
+    df["(T2) Entailment (TRUE %)"] = df["(T2) Entailment (TRUE %)"].apply(extract_first)
+    df["(T2) Entailment (GPT-4o %)"] = df["(T2) Entailment (GPT-4o %)"].apply(extract_first)
+    df["(T2) ROUGE-1 (%)"] = df["(T2) ROUGE-1 (%)"].apply(extract_first)
+    df["(T2) ROUGE-2 (%)"] = df["(T2) ROUGE-2 (%)"].apply(extract_first)
+    df["(T2) ROUGE-L (%)"] = df["(T2) ROUGE-L (%)"].apply(extract_first)
     # Task 3
-    df["(T3) Precision (%)"] = df["(T3) Precision (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T3) Title_search_rate (%)"] = df["(T3) Title_search_rate (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T3) Overlap (%)"] = df["(T3) Overlap (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T3) Key Point Recall (%)"] = df["(T3) Key Point Recall (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T3) ROUGE-1 (%)"] = df["(T3) ROUGE-1 (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T3) ROUGE-2 (%)"] = df["(T3) ROUGE-2 (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
-    df["(T3) ROUGE-L (%)"] = df["(T3) ROUGE-L (%)"].apply(lambda x: x[0] if len(x) > 0 else 0)
+    df["(T3) Precision (%)"] = df["(T3) Precision (%)"].apply(extract_first)
+    df["(T3) Title_search_rate (%)"] = df["(T3) Title_search_rate (%)"].apply(extract_first)
+    df["(T3) Overlap (%)"] = df["(T3) Overlap (%)"].apply(extract_first)
+    df["(T3) Key Point Recall (%)"] = df["(T3) Key Point Recall (%)"].apply(extract_first)
+    df["(T3) ROUGE-1 (%)"] = df["(T3) ROUGE-1 (%)"].apply(extract_first)
+    df["(T3) ROUGE-2 (%)"] = df["(T3) ROUGE-2 (%)"].apply(extract_first)
+    df["(T3) ROUGE-L (%)"] = df["(T3) ROUGE-L (%)"].apply(extract_first)
     
     # 平均值列
     df["Average ⬆️"] = df[["(T1) Precision (%)",