update

src/leaderboard/read_evals.py

@@ -73,11 +73,13 @@ class EvalResult:
 
             # We average all scores of a given metric (not all metrics are present in all files)
             accs = np.array([v.get(task.metric, None) for k, v in data["results"].items() if task.benchmark == k])
-            # if accs.size == 0 or any([acc is None for acc in accs]):
+            if accs.size == 0 or any([acc is None for acc in accs]):
+                mean_acc = 0.0
+            else:
+                mean_acc = np.nanmean(accs) * 100.0
             # if accs.size == 0:
             #     continue
             print(accs)
-            mean_acc = np.nanmean(accs) * 100.0
             # mean_acc = np.mean(accs) * 100.0
             results[task.benchmark] = mean_acc