Bigbench fix (#1686)

* edit process multiple-choice

* split template yaml

* remove

* modified multiple_choice tasks

* udpate

* Update multiple_choice_template_b_yaml

* Update multiple_choice_template_a_yaml

---------
Co-authored-by: Hailey Schoelkopf <65563625+haileyschoelkopf@users.noreply.github.com>

lm_eval/tasks/bigbench/multiple_choice/unit_conversion.yaml

 # Generated by utils.py
 dataset_name: unit_conversion_zero_shot
-include: ../multiple_choice_template_yaml
+include: ../multiple_choice_template_a_yaml
 task: bigbench_unit_conversion_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/unit_interpretation.yaml

 # Generated by utils.py
 dataset_name: unit_interpretation_zero_shot
-include: ../multiple_choice_template_yaml
+include: ../multiple_choice_template_a_yaml
 task: bigbench_unit_interpretation_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/unnatural_in_context_learning.yaml

-# Generated by utils.py
-dataset_name: unnatural_in_context_learning_zero_shot
-include: ../multiple_choice_template_yaml
-task: bigbench_unnatural_in_context_learning_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/vitaminc_fact_verification.yaml

 # Generated by utils.py
 dataset_name: vitaminc_fact_verification_zero_shot
-include: ../multiple_choice_template_yaml
+include: ../multiple_choice_template_a_yaml
 task: bigbench_vitaminc_fact_verification_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/what_is_the_tao.yaml

 # Generated by utils.py
 dataset_name: what_is_the_tao_zero_shot
-include: ../multiple_choice_template_yaml
+include: ../multiple_choice_template_a_yaml
 task: bigbench_what_is_the_tao_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/which_wiki_edit.yaml

 # Generated by utils.py
 dataset_name: which_wiki_edit_zero_shot
-include: ../multiple_choice_template_yaml
+include: ../multiple_choice_template_a_yaml
 task: bigbench_which_wiki_edit_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/winowhy.yaml

 # Generated by utils.py
 dataset_name: winowhy_zero_shot
-include: ../multiple_choice_template_yaml
+include: ../multiple_choice_template_a_yaml
 task: bigbench_winowhy_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/word_sorting.yaml

-# Generated by utils.py
-dataset_name: word_sorting_zero_shot
-include: ../multiple_choice_template_yaml
-task: bigbench_word_sorting_multiple_choice

lm_eval/tasks/bigbench/multiple_choice/word_unscrambling.yaml

-# Generated by utils.py
-dataset_name: word_unscrambling_zero_shot
-include: ../multiple_choice_template_yaml
-task: bigbench_word_unscrambling_multiple_choice

lm_eval/tasks/bigbench/multiple_choice_template_yaml → lm_eval/tasks/bigbench/multiple_choice_template_a_yaml

@@ -12,4 +12,4 @@ metric_list:
   - metric: acc
   # TODO: brier score and other metrics
 metadata:
-  version: 0.0
+  version: 1.0

lm_eval/tasks/bigbench/multiple_choice_template_b_yaml

+group: bigbench_multiple_choice
+dataset_path: hails/bigbench
+dataset_kwargs:
+  # num_shots: 0 # TODO: num of shots for `bigbench` HF dataset should be controlled through this, not through the typical methods
+  # subtask_name: null
+output_type: multiple_choice
+test_split: default
+doc_to_text: inputs
+doc_to_target: "{{multiple_choice_scores.index(1)}}"
+doc_to_choice: "{{multiple_choice_targets}}"
+metric_list:
+  - metric: acc
+  # TODO: brier score and other metrics
+metadata:
+  version: 1.0