brier score is working for N-sized class

lm_eval/api/metrics.py

@@ -109,12 +109,10 @@ def ter(items):
 
 @register_aggregation("brier_score")
 def brier_score(items):  # This is a passthrough function
-    gold = list(zip(*items))[0]
-    gold_one_hot = np.eye(max(gold)+1)[gold]
+    gold, predictions = list(zip(*items))
+    gold = list(gold)
+    gold_one_hot = np.eye(np.max(gold)+1)[gold]
     predictions = list(zip(*items))[1]
-    print("predictions", prediction)
-    print("gold_one_hot", gold_one_hot)
-    import sys; sys.exit()
     return np.mean(np.sum((predictions - gold_one_hot)**2, axis=1))