add comma bench

lm_eval/tasks/benchmarks/comma.yaml

+group: comma
+task:
+  - task: arc_challenge
+    metric_list:
+      - metric: acc_mutual_info
+        aggregation: mean
+        higher_is_better: true
+  - task: arc_easy
+    metric_list:
+      - metric: acc_mutual_info
+        aggregation: mean
+        higher_is_better: true
+  - boolq
+  - task: hellaswag
+    metric_list:
+      - metric: acc_norm
+        aggregation: mean
+        higher_is_better: true
+  - task: openbookqa
+    metric_list:
+      - metric: acc_mutual_info
+        aggregation: mean
+        higher_is_better: true
+  - task: commonsense_qa
+    doc_to_text: "Question: {{ question.strip() }}\nAnswer:"
+    doc_to_target: '{{["A", "B", "C", "D", "E"].index(answerKey)}}'
+    doc_to_choice: "{{ choices['text'] }}"
+    metric_list:
+      - metric: acc_mutual_info
+        aggregation: mean
+        higher_is_better: true
+  - task: piqa
+    doc_to_text: "Goal: {{goal}}\nAnswer:"
+    metric_list:
+      - metric: acc_norm
+        aggregation: mean
+        higher_is_better: true
+  - task: social_iqa
+    doc_to_text: "Question: {{context}} {{question}}\nAnswer:"
+    metric_list:
+      - metric: acc_norm
+        aggregation: mean
+        higher_is_better: true
+aggregate_metric_list:
+  - metric: acc
+    aggregation: mean
+    weight_by_size: false
+  - metric: acc_norm
+    aggregation: mean
+    weight_by_size: false
+  - metric: acc_mutual_info
+    aggregation: mean
+    weight_by_size: false

lm_eval/tasks/super_glue/boolq/default.yaml

@@ -13,5 +13,7 @@ should_decontaminate: true
 doc_to_decontamination_query: passage
 metric_list:
   - metric: acc
+    aggregation: mean
+    higher_is_better: true
 metadata:
   version: 2.0