update

src/leaderboard/read_evals.py

@@ -89,10 +89,18 @@ class EvalResult:
 
             def post_process_results(results: dict) -> dict:
                 res_copy = results.copy()
+
+                for k, v in res_copy.items():
+                    if "exact_match" in k:
+                        results[k.replace("exact_match", "em")] = v
+
+                res_copy = results.copy()
+
                 for k, v in res_copy.items():
                     if "," in k:
                         tokens = k.split(",")
                         results[tokens[0]] = v
+
                 return results
 
             accs = np.array([v.get(task.metric, None) for k, v in post_process_results(data["results"]).items() if task.benchmark in k])