auroc does not compute CI by default from now on.

DCGM · Sep 19, 2024 · c5f56c3 · c5f56c3
1 parent 72b8c5d
commit c5f56c3
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/lm_eval/tasks/benczechmark/utils.py b/lm_eval/tasks/benczechmark/utils.py
@@ -220,7 +220,7 @@ def aggregate_CI_avg_mcauroc(items):
     probs = unzipped_list[1]
 
     metric = evaluate.load("CZLC/mc_auroc")
-    result = metric.compute(predictions=probs, references=golds)
+    result = metric.compute(predictions=probs, references=golds, CI=True)
     return result["mc_auroc_ci"]