merged with latest update

lm_eval/tasks/bbh/cot_zeroshot/disambiguation_qa.yaml

 "dataset_name": "disambiguation_qa"
 "description": "Clarify the meaning of sentences with ambiguous pronouns.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_disambiguation_qa"
+
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/dyck_languages.yaml

 "dataset_name": "dyck_languages"
 "description": "Correctly close a Dyck-n word.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_dyck_languages"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: "regex"
+        group_select: -1
+        regex_pattern: "(?<= )([\" \\[\\(<{}>\\)\\]]+)|([\" \\[\\(<{}>\\)\\]]+)"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/formal_fallacies.yaml

 "dataset_name": "formal_fallacies"
 "description": "Distinguish deductively valid arguments from formal fallacies.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_formal_fallacies"
+
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: "regex"
+        group_select: -1
+        regex_pattern: "\\b(valid|invalid)\\b"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/geometric_shapes.yaml

 "dataset_name": "geometric_shapes"
 "description": "Name geometric shapes from their SVG paths.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_geometric_shapes"
+
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/hyperbaton.yaml

 "dataset_name": "hyperbaton"
 "description": "Order adjectives correctly in English sentences.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_hyperbaton"
+
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/logical_deduction_five_objects.yaml

 "dataset_name": "logical_deduction_five_objects"
 "description": "A logical deduction task which requires deducing the order of a sequence of objects.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_logical_deduction_five_objects"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/logical_deduction_seven_objects.yaml

 "dataset_name": "logical_deduction_seven_objects"
 "description": "A logical deduction task which requires deducing the order of a sequence of objects.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_logical_deduction_seven_objects"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/logical_deduction_three_objects.yaml

 "dataset_name": "logical_deduction_three_objects"
 "description": "A logical deduction task which requires deducing the order of a sequence of objects.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_logical_deduction_three_objects"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/movie_recommendation.yaml

 "dataset_name": "movie_recommendation"
 "description": "Recommend movies similar to the given list of movies.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_movie_recommendation"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/multistep_arithmetic_two.yaml

 "dataset_name": "multistep_arithmetic_two"
 "description": "Solve multi-step arithmetic problems.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_multistep_arithmetic_two"
+
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.NumberParseRegexFilter
+        group_select: -1
+        regex_pattern: "([-0-9]+)"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"
+

lm_eval/tasks/bbh/cot_zeroshot/navigate.yaml

 "dataset_name": "navigate"
 "description": "Given a series of navigation instructions, determine whether one would end up back at the starting point.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_navigate"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: "regex"
+        group_select: -1
+        regex_pattern: "\\b(Yes|No|yes|no)\\b"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/object_counting.yaml

 "dataset_name": "object_counting"
 "description": "Questions that involve enumerating objects and asking the model to count them.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_object_counting"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.NumberParseRegexFilter
+        group_select: -1
+        regex_pattern: "([-0-9]+)"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"
+

lm_eval/tasks/bbh/cot_zeroshot/penguins_in_a_table.yaml

 "dataset_name": "penguins_in_a_table"
 "description": "Answer questions about a table of penguins and their attributes.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_penguins_in_a_table"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/reasoning_about_colored_objects.yaml

 "dataset_name": "reasoning_about_colored_objects"
 "description": "Answer extremely simple questions about the colors of objects on a surface.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_reasoning_about_colored_objects"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/ruin_names.yaml

 "dataset_name": "ruin_names"
 "description": "Select the humorous edit that 'ruins' the input movie or musical artist name.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_ruin_names"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/salient_translation_error_detection.yaml

 "dataset_name": "salient_translation_error_detection"
 "description": "Detect the type of error in an English translation of a German source sentence.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_salient_translation_error_detection"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/snarks.yaml

 "dataset_name": "snarks"
 "description": "Determine which of two sentences is sarcastic.\n\nAccording to Cambridge University Dictionary, sarcasm is \"the use of remarks that clearly mean the opposite of what they say, made in order to hurt someone's feelings or to criticize something in a humorous way.\" Sarcastic sentences often contain satirical or ironic utterances, hyperboles, ambivalent or witty remarks.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_snarks"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/sports_understanding.yaml

 "dataset_name": "sports_understanding"
 "description": "Determine whether an artificially constructed sentence relating to sports is plausible or not.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_sports_understanding"
+
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MapRegexFilter
+        group_select: -1
+        ignore_case: true
+        regex_pattern_to_value:
+          \b(no|not plausible)\b: "no"
+          \b(yes|plausible)\b: "yes"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/temporal_sequences.yaml

 "dataset_name": "temporal_sequences"
 "description": "Task description: Answer questions about which times certain events could have occurred.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_temporal_sequences"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"

lm_eval/tasks/bbh/cot_zeroshot/tracking_shuffled_objects_five_objects.yaml

 "dataset_name": "tracking_shuffled_objects_five_objects"
 "description": "A task requiring determining the final positions of a set of objects given their initial positions and a description of a sequence of swaps.\n\n"
-"doc_to_text": "Q: {{input}}\nA: Let's think step by step.\n"
+"doc_to_text": "Q: {{input}}\nA: Let's think step by step."
 "include": "_cot_zeroshot_template_yaml"
 "task": "bbh_cot_zeroshot_tracking_shuffled_objects_five_objects"
+filter_list:
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: -1
+        ignore_case: true
+        ignore_punctuation: true
+        regex_pattern: "(\\([A-Z]\\))"
+      - function: "take_first"
+  - name: "strict-match"
+    filter:
+      - function: "regex"
+        regex_pattern: "((?<=The answer is )(.*)(?=.)|(?<=the answer is )(.*)(?=.)|(?<=The answer: )(.*)(?=.)|(?<=The final answer: )(.*)(?=.))"
+      - function: "take_first"