Rolled back gsm8k metrics

lm_eval/tasks/gsm8k/gsm8k-ru.yaml

@@ -9,5 +9,33 @@ generation_kwargs:
   until:
     - "\n\n"
     - "Вопрос:"
+    - "</s>"
+    - "<|im_end|>"
   do_sample: false
   temperature: 0.0
+output_type: generate_until
+training_split: train
+fewshot_split: train
+test_split: test
+doc_to_target: "{{answer}}" #" {{answer.split('### ')[-1].rstrip()}}"
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+    ignore_case: true
+    ignore_punctuation: false
+    regexes_to_ignore:
+      - ","
+      - "\\$"
+      - "(?s).*#### "
+      - "\\.$"
+repeats: 1
+num_fewshot: 5
+filter_list:
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "#### (\\-?[0-9\\.\\,]+)"
+      - function: "take_first"
+metadata:
+  version: 3.0