Add files using upload-large-folder tool

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/_mmlu_flan_generative_template_yaml

@@ -0,0 +1,34 @@
+dataset_path: hails/mmlu_no_train # a copy of `cais/mmlu` with no auxiliary_train split
+test_split: test
+fewshot_split: dev
+fewshot_config:
+  sampler: first_n
+output_type: generate_until
+doc_to_text: "Q: {{question.strip()}}\n(A) {{choices[0]}} (B) {{choices[1]}} (C) {{choices[2]}} (D) {{choices[3]}}\nA:"
+doc_to_target: "{{['(A)', '(B)', '(C)', '(D)'][answer]}}"
+filter_list:
+  - name: "strict-match"
+    filter:
+      - function: "take_first"
+  - name: "flexible-extract"
+    filter:
+      - function: !function utils.MultiChoiceRegexFilter
+        group_select: 0
+        regex_pattern: "(\\([A-Z]\\))"
+        ignore_case: true
+        ignore_punctuation: true
+      - function: "take_first"
+generation_kwargs:
+  until:
+    - "</s>"
+    - "Q:"
+    - "<|im_end|>"
+    - "\n"
+metric_list:
+  - metric: exact_match
+    aggregation: mean
+    higher_is_better: true
+metadata:
+  version: 3.0
+dataset_kwargs:
+  trust_remote_code: true

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_anatomy.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "anatomy"
+"description": "The following are multiple choice questions (with answers) about anatomy.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_anatomy"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_astronomy.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "astronomy"
+"description": "The following are multiple choice questions (with answers) about astronomy.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_astronomy"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_college_medicine.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "college_medicine"
+"description": "The following are multiple choice questions (with answers) about college\
+  \ medicine.\n\n"
+"tag": "mmlu_flan_n_shot_generative_other"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_college_medicine"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_computer_security.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "computer_security"
+"description": "The following are multiple choice questions (with answers) about computer\
+  \ security.\n\n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_computer_security"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_conceptual_physics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "conceptual_physics"
+"description": "The following are multiple choice questions (with answers) about conceptual\
+  \ physics.\n\n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_conceptual_physics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_econometrics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "econometrics"
+"description": "The following are multiple choice questions (with answers) about econometrics.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_social_sciences"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_econometrics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_electrical_engineering.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "electrical_engineering"
+"description": "The following are multiple choice questions (with answers) about electrical\
+  \ engineering.\n\n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_electrical_engineering"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_formal_logic.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "formal_logic"
+"description": "The following are multiple choice questions (with answers) about formal\
+  \ logic.\n\n"
+"tag": "mmlu_flan_n_shot_generative_humanities"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_formal_logic"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_global_facts.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "global_facts"
+"description": "The following are multiple choice questions (with answers) about global\
+  \ facts.\n\n"
+"tag": "mmlu_flan_n_shot_generative_other"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_global_facts"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_high_school_chemistry.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_chemistry"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school chemistry.\n\n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_high_school_chemistry"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_high_school_mathematics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_mathematics"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school mathematics.\n\n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_high_school_mathematics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_high_school_microeconomics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_microeconomics"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school microeconomics.\n\n"
+"tag": "mmlu_flan_n_shot_generative_social_sciences"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_high_school_microeconomics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_high_school_psychology.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_psychology"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school psychology.\n\n"
+"tag": "mmlu_flan_n_shot_generative_social_sciences"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_high_school_psychology"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_high_school_statistics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_statistics"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school statistics.\n\n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_high_school_statistics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_international_law.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "international_law"
+"description": "The following are multiple choice questions (with answers) about international\
+  \ law.\n\n"
+"tag": "mmlu_flan_n_shot_generative_humanities"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_international_law"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_jurisprudence.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "jurisprudence"
+"description": "The following are multiple choice questions (with answers) about jurisprudence.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_humanities"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_jurisprudence"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_machine_learning.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "machine_learning"
+"description": "The following are multiple choice questions (with answers) about machine\
+  \ learning.\n\n"
+"tag": "mmlu_flan_n_shot_generative_stem"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_machine_learning"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_management.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "management"
+"description": "The following are multiple choice questions (with answers) about management.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_other"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_management"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_marketing.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "marketing"
+"description": "The following are multiple choice questions (with answers) about marketing.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_other"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_marketing"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_medical_genetics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "medical_genetics"
+"description": "The following are multiple choice questions (with answers) about medical\
+  \ genetics.\n\n"
+"tag": "mmlu_flan_n_shot_generative_other"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_medical_genetics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_miscellaneous.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "miscellaneous"
+"description": "The following are multiple choice questions (with answers) about miscellaneous.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_other"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_miscellaneous"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_moral_scenarios.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "moral_scenarios"
+"description": "The following are multiple choice questions (with answers) about moral\
+  \ scenarios.\n\n"
+"tag": "mmlu_flan_n_shot_generative_humanities"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_moral_scenarios"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_nutrition.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "nutrition"
+"description": "The following are multiple choice questions (with answers) about nutrition.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_other"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_nutrition"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_prehistory.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "prehistory"
+"description": "The following are multiple choice questions (with answers) about prehistory.\n\
+  \n"
+"tag": "mmlu_flan_n_shot_generative_humanities"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_prehistory"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_professional_psychology.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "professional_psychology"
+"description": "The following are multiple choice questions (with answers) about professional\
+  \ psychology.\n\n"
+"tag": "mmlu_flan_n_shot_generative_social_sciences"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_professional_psychology"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_security_studies.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "security_studies"
+"description": "The following are multiple choice questions (with answers) about security\
+  \ studies.\n\n"
+"tag": "mmlu_flan_n_shot_generative_social_sciences"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_security_studies"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_us_foreign_policy.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "us_foreign_policy"
+"description": "The following are multiple choice questions (with answers) about us\
+  \ foreign policy.\n\n"
+"tag": "mmlu_flan_n_shot_generative_social_sciences"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_us_foreign_policy"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/generative/mmlu_world_religions.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "world_religions"
+"description": "The following are multiple choice questions (with answers) about world\
+  \ religions.\n\n"
+"tag": "mmlu_flan_n_shot_generative_humanities"
+"include": "_mmlu_flan_generative_template_yaml"
+"task": "mmlu_flan_n_shot_generative_world_religions"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/_mmlu.yaml

@@ -0,0 +1,32 @@
+group: mmlu_flan_n_shot_loglikelihood
+group_alias: mmlu (flan style, loglikelihood)
+task:
+  - group: stem
+    task:
+      - mmlu_flan_n_shot_loglikelihood_stem
+    aggregate_metric_list:
+      - metric: acc
+        weight_by_size: True
+  - group: other
+    task:
+      - mmlu_flan_n_shot_loglikelihood_other
+    aggregate_metric_list:
+      - metric: acc
+        weight_by_size: True
+  - group: social sciences
+    task:
+      - mmlu_flan_n_shot_loglikelihood_social_sciences
+    aggregate_metric_list:
+      - metric: acc
+        weight_by_size: True
+  - group: humanities
+    task:
+      - mmlu_flan_n_shot_loglikelihood_humanities
+    aggregate_metric_list:
+      - metric: acc
+        weight_by_size: True
+aggregate_metric_list:
+  - metric: acc
+    weight_by_size: True
+metadata:
+  version: 2

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/_mmlu_flan_loglikelihood_template_yaml

@@ -0,0 +1,17 @@
+dataset_path: hails/mmlu_no_train # a copy of `cais/mmlu` with no auxiliary_train split
+test_split: test
+fewshot_split: dev
+fewshot_config:
+  sampler: first_n
+output_type: multiple_choice
+doc_to_text: "Q: {{question.strip()}}\n(A) {{choices[0]}} (B) {{choices[1]}} (C) {{choices[2]}} (D) {{choices[3]}}\nA:"
+doc_to_choice: ["(A)", "(B)", "(C)", "(D)"]
+doc_to_target: answer
+metric_list:
+  - metric: acc
+    aggregation: mean
+    higher_is_better: true
+metadata:
+  version: 2.0
+dataset_kwargs:
+  trust_remote_code: true

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_abstract_algebra.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "abstract_algebra"
+"description": "The following are multiple choice questions (with answers) about abstract\
+  \ algebra.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_abstract_algebra"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_business_ethics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "business_ethics"
+"description": "The following are multiple choice questions (with answers) about business\
+  \ ethics.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_other"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_business_ethics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_clinical_knowledge.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "clinical_knowledge"
+"description": "The following are multiple choice questions (with answers) about clinical\
+  \ knowledge.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_other"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_clinical_knowledge"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_college_biology.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "college_biology"
+"description": "The following are multiple choice questions (with answers) about college\
+  \ biology.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_college_biology"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_college_chemistry.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "college_chemistry"
+"description": "The following are multiple choice questions (with answers) about college\
+  \ chemistry.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_college_chemistry"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_college_computer_science.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "college_computer_science"
+"description": "The following are multiple choice questions (with answers) about college\
+  \ computer science.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_college_computer_science"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_college_mathematics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "college_mathematics"
+"description": "The following are multiple choice questions (with answers) about college\
+  \ mathematics.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_college_mathematics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_college_physics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "college_physics"
+"description": "The following are multiple choice questions (with answers) about college\
+  \ physics.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_college_physics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_conceptual_physics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "conceptual_physics"
+"description": "The following are multiple choice questions (with answers) about conceptual\
+  \ physics.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_conceptual_physics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_electrical_engineering.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "electrical_engineering"
+"description": "The following are multiple choice questions (with answers) about electrical\
+  \ engineering.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_electrical_engineering"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_elementary_mathematics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "elementary_mathematics"
+"description": "The following are multiple choice questions (with answers) about elementary\
+  \ mathematics.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_elementary_mathematics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_global_facts.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "global_facts"
+"description": "The following are multiple choice questions (with answers) about global\
+  \ facts.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_other"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_global_facts"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_high_school_biology.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_biology"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school biology.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_high_school_biology"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_high_school_chemistry.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_chemistry"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school chemistry.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_high_school_chemistry"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_high_school_computer_science.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_computer_science"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school computer science.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_stem"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_high_school_computer_science"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_high_school_european_history.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_european_history"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school european history.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_humanities"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_high_school_european_history"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_high_school_geography.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_geography"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school geography.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_social_sciences"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_high_school_geography"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_high_school_government_and_politics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_government_and_politics"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school government and politics.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_social_sciences"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_high_school_government_and_politics"

scripts/yans/lm-evaluation-harness/lm_eval/tasks/mmlu/flan_n_shot/loglikelihood/mmlu_high_school_macroeconomics.yaml

@@ -0,0 +1,6 @@
+"dataset_name": "high_school_macroeconomics"
+"description": "The following are multiple choice questions (with answers) about high\
+  \ school macroeconomics.\n\n"
+"tag": "mmlu_flan_n_shot_loglikelihood_social_sciences"
+"include": "_mmlu_flan_loglikelihood_template_yaml"
+"task": "mmlu_flan_n_shot_loglikelihood_high_school_macroeconomics"