diff --git "a/evaluations/ar/Allam-7b-instruct-preview/arabicmmlu_0_shot.json" "b/evaluations/ar/Allam-7b-instruct-preview/arabicmmlu_0_shot.json" new file mode 100644--- /dev/null +++ "b/evaluations/ar/Allam-7b-instruct-preview/arabicmmlu_0_shot.json" @@ -0,0 +1,2086 @@ +{ + "results": { + "arabicmmlu": { + "acc,none": 0.6777585610515393, + "acc_stderr,none": 0.0037651094938210825, + "alias": "arabicmmlu" + }, + "arabicmmlu_humanities": { + "acc,none": 0.7196802646085998, + "acc_stderr,none": 0.007156852970625745, + "alias": " - Humanities" + }, + "arabicmmlu_high_history": { + "alias": " - High History", + "acc,none": 0.5039473684210526, + "acc_stderr,none": 0.01814828462669052 + }, + "arabicmmlu_high_islamic_studies": { + "alias": " - High Islamic Studies", + "acc,none": 0.7485029940119761, + "acc_stderr,none": 0.023776124368602287 + }, + "arabicmmlu_high_philosophy": { + "alias": " - High Philosophy", + "acc,none": 0.7435897435897436, + "acc_stderr,none": 0.07083413480167725 + }, + "arabicmmlu_islamic_studies": { + "alias": " - Islamic Studies", + "acc,none": 0.704225352112676, + "acc_stderr,none": 0.018068660651366884 + }, + "arabicmmlu_middle_history": { + "alias": " - Middle History", + "acc,none": 0.7241379310344828, + "acc_stderr,none": 0.03144712581678242 + }, + "arabicmmlu_middle_islamic_studies": { + "alias": " - Middle Islamic Studies", + "acc,none": 0.7647058823529411, + "acc_stderr,none": 0.027553614467863807 + }, + "arabicmmlu_primary_history": { + "alias": " - Primary History", + "acc,none": 0.7647058823529411, + "acc_stderr,none": 0.04220773659171455 + }, + "arabicmmlu_primary_islamic_studies": { + "alias": " - Primary Islamic Studies", + "acc,none": 0.8708708708708709, + "acc_stderr,none": 0.010615091024310195 + }, + "arabicmmlu_prof_law": { + "alias": " - Prof Law", + "acc,none": 0.7070063694267515, + "acc_stderr,none": 0.025725781937262132 + }, + "arabicmmlu_language": { + "acc,none": 0.7053462940461726, + "acc_stderr,none": 0.010675632352174308, + "alias": " - Language" + }, + "arabicmmlu_arabic_language_(general)": { + "alias": " - Arabic Language (General)", + "acc,none": 0.8088235294117647, + "acc_stderr,none": 0.01590829013627805 + }, + "arabicmmlu_arabic_language_(grammar)": { + "alias": " - Arabic Language (Grammar)", + "acc,none": 0.7232876712328767, + "acc_stderr,none": 0.02344871747678411 + }, + "arabicmmlu_high_arabic_language": { + "alias": " - High Arabic Language", + "acc,none": 0.45384615384615384, + "acc_stderr,none": 0.025242770987126177 + }, + "arabicmmlu_middle_arabic_language": { + "alias": " - Middle Arabic Language", + "acc,none": 0.8518518518518519, + "acc_stderr,none": 0.06966962541673782 + }, + "arabicmmlu_primary_arabic_language": { + "alias": " - Primary Arabic Language", + "acc,none": 0.8015873015873016, + "acc_stderr,none": 0.025172322396351483 + }, + "arabicmmlu_other": { + "acc,none": 0.7089371980676329, + "acc_stderr,none": 0.009115340366470213, + "alias": " - Other" + }, + "arabicmmlu_driving_test": { + "alias": " - Driving Test", + "acc,none": 0.6985962014863749, + "acc_stderr,none": 0.013191518335507111 + }, + "arabicmmlu_general_knowledge": { + "alias": " - General Knowledge", + "acc,none": 0.7199074074074074, + "acc_stderr,none": 0.015285643798521893 + }, + "arabicmmlu_middle_general_knowledge": { + "alias": " - Middle General Knowledge", + "acc,none": 0.6802325581395349, + "acc_stderr,none": 0.035665455380848116 + }, + "arabicmmlu_primary_general_knowledge": { + "alias": " - Primary General Knowledge", + "acc,none": 0.7654320987654321, + "acc_stderr,none": 0.03339448023577033 + }, + "arabicmmlu_univ_management": { + "alias": " - Univ Management", + "acc,none": 0.6933333333333334, + "acc_stderr,none": 0.05360292224565066 + }, + "arabicmmlu_social_science": { + "acc,none": 0.641837899543379, + "acc_stderr,none": 0.00797908211240422, + "alias": " - Social Science" + }, + "arabicmmlu_high_civics": { + "alias": " - High Civics", + "acc,none": 0.4827586206896552, + "acc_stderr,none": 0.05388432214060092 + }, + "arabicmmlu_high_economics": { + "alias": " - High Economics", + "acc,none": 0.625, + "acc_stderr,none": 0.025551030374592384 + }, + "arabicmmlu_high_geography": { + "alias": " - High Geography", + "acc,none": 0.5770712909441233, + "acc_stderr,none": 0.015341186146893518 + }, + "arabicmmlu_middle_civics": { + "alias": " - Middle Civics", + "acc,none": 0.5932203389830508, + "acc_stderr,none": 0.03204451480926517 + }, + "arabicmmlu_middle_economics": { + "alias": " - Middle Economics", + "acc,none": 0.7471264367816092, + "acc_stderr,none": 0.04687049503854671 + }, + "arabicmmlu_middle_geography": { + "alias": " - Middle Geography", + "acc,none": 0.7132352941176471, + "acc_stderr,none": 0.02747227447323382 + }, + "arabicmmlu_middle_social_science": { + "alias": " - Middle Social Science", + "acc,none": 0.5767634854771784, + "acc_stderr,none": 0.03189222523446444 + }, + "arabicmmlu_primary_geography": { + "alias": " - Primary Geography", + "acc,none": 0.7719298245614035, + "acc_stderr,none": 0.05606981784761176 + }, + "arabicmmlu_primary_social_science": { + "alias": " - Primary Social Science", + "acc,none": 0.7815602836879433, + "acc_stderr,none": 0.015572585115281092 + }, + "arabicmmlu_univ_accounting": { + "alias": " - Univ Accounting", + "acc,none": 0.6351351351351351, + "acc_stderr,none": 0.05634270081349515 + }, + "arabicmmlu_univ_economics": { + "alias": " - Univ Economics", + "acc,none": 0.5693430656934306, + "acc_stderr,none": 0.04246032224326305 + }, + "arabicmmlu_univ_political_science": { + "alias": " - Univ Political Science", + "acc,none": 0.5952380952380952, + "acc_stderr,none": 0.03395252139627751 + }, + "arabicmmlu_stem": { + "acc,none": 0.6310679611650486, + "acc_stderr,none": 0.008195409873199793, + "alias": " - STEM" + }, + "arabicmmlu_high_biology": { + "alias": " - High Biology", + "acc,none": 0.5095812633073101, + "acc_stderr,none": 0.013322598053209577 + }, + "arabicmmlu_high_computer_science": { + "alias": " - High Computer Science", + "acc,none": 0.6934865900383141, + "acc_stderr,none": 0.02859282719866765 + }, + "arabicmmlu_high_physics": { + "alias": " - High Physics", + "acc,none": 0.5176470588235295, + "acc_stderr,none": 0.031353244021767535 + }, + "arabicmmlu_middle_computer_science": { + "alias": " - Middle Computer Science", + "acc,none": 0.9259259259259259, + "acc_stderr,none": 0.051361129280113826 + }, + "arabicmmlu_middle_natural_science": { + "alias": " - Middle Natural Science", + "acc,none": 0.8016528925619835, + "acc_stderr,none": 0.02568606613318377 + }, + "arabicmmlu_primary_computer_science": { + "alias": " - Primary Computer Science", + "acc,none": 0.7473684210526316, + "acc_stderr,none": 0.031606782497111685 + }, + "arabicmmlu_primary_math": { + "alias": " - Primary Math", + "acc,none": 0.6772616136919315, + "acc_stderr,none": 0.023145867389961022 + }, + "arabicmmlu_primary_natural_science": { + "alias": " - Primary Natural Science", + "acc,none": 0.8839285714285714, + "acc_stderr,none": 0.017500435136664095 + }, + "arabicmmlu_univ_computer_science": { + "alias": " - Univ Computer Science", + "acc,none": 0.765625, + "acc_stderr,none": 0.053369535239372906 + } + }, + "groups": { + "arabicmmlu": { + "acc,none": 0.6777585610515393, + "acc_stderr,none": 0.0037651094938210825, + "alias": "arabicmmlu" + }, + "arabicmmlu_humanities": { + "acc,none": 0.7196802646085998, + "acc_stderr,none": 0.007156852970625745, + "alias": " - Humanities" + }, + "arabicmmlu_language": { + "acc,none": 0.7053462940461726, + "acc_stderr,none": 0.010675632352174308, + "alias": " - Language" + }, + "arabicmmlu_other": { + "acc,none": 0.7089371980676329, + "acc_stderr,none": 0.009115340366470213, + "alias": " - Other" + }, + "arabicmmlu_social_science": { + "acc,none": 0.641837899543379, + "acc_stderr,none": 0.00797908211240422, + "alias": " - Social Science" + }, + "arabicmmlu_stem": { + "acc,none": 0.6310679611650486, + "acc_stderr,none": 0.008195409873199793, + "alias": " - STEM" + } + }, + "group_subtasks": { + "arabicmmlu_language": [ + "arabicmmlu_high_arabic_language", + "arabicmmlu_arabic_language_(grammar)", + "arabicmmlu_middle_arabic_language", + "arabicmmlu_arabic_language_(general)", + "arabicmmlu_primary_arabic_language" + ], + "arabicmmlu_stem": [ + "arabicmmlu_high_computer_science", + "arabicmmlu_primary_math", + "arabicmmlu_high_biology", + "arabicmmlu_primary_computer_science", + "arabicmmlu_middle_natural_science", + "arabicmmlu_high_physics", + "arabicmmlu_middle_computer_science", + "arabicmmlu_univ_computer_science", + "arabicmmlu_primary_natural_science" + ], + "arabicmmlu_humanities": [ + "arabicmmlu_prof_law", + "arabicmmlu_middle_islamic_studies", + "arabicmmlu_high_philosophy", + "arabicmmlu_high_islamic_studies", + "arabicmmlu_islamic_studies", + "arabicmmlu_high_history", + "arabicmmlu_primary_islamic_studies", + "arabicmmlu_middle_history", + "arabicmmlu_primary_history" + ], + "arabicmmlu_social_science": [ + "arabicmmlu_middle_social_science", + "arabicmmlu_univ_economics", + "arabicmmlu_univ_accounting", + "arabicmmlu_high_civics", + "arabicmmlu_high_economics", + "arabicmmlu_middle_geography", + "arabicmmlu_primary_geography", + "arabicmmlu_middle_civics", + "arabicmmlu_high_geography", + "arabicmmlu_middle_economics", + "arabicmmlu_univ_political_science", + "arabicmmlu_primary_social_science" + ], + "arabicmmlu_other": [ + "arabicmmlu_primary_general_knowledge", + "arabicmmlu_general_knowledge", + "arabicmmlu_middle_general_knowledge", + "arabicmmlu_univ_management", + "arabicmmlu_driving_test" + ], + "arabicmmlu": [ + "arabicmmlu_other", + "arabicmmlu_social_science", + "arabicmmlu_humanities", + "arabicmmlu_stem", + "arabicmmlu_language" + ] + }, + "configs": { + "arabicmmlu_arabic_language_(general)": { + "task": "arabicmmlu_arabic_language_(general)", + "task_alias": "Arabic Language (General)", + "tag": "arabicmmlu_language_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Arabic Language (General)", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_arabic_language_(grammar)": { + "task": "arabicmmlu_arabic_language_(grammar)", + "task_alias": "Arabic Language (Grammar)", + "tag": "arabicmmlu_language_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Arabic Language (Grammar)", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_driving_test": { + "task": "arabicmmlu_driving_test", + "task_alias": "Driving Test", + "tag": "arabicmmlu_other_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Driving Test", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_general_knowledge": { + "task": "arabicmmlu_general_knowledge", + "task_alias": "General Knowledge", + "tag": "arabicmmlu_other_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "General Knowledge", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_arabic_language": { + "task": "arabicmmlu_high_arabic_language", + "task_alias": "High Arabic Language", + "tag": "arabicmmlu_language_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Arabic Language", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_biology": { + "task": "arabicmmlu_high_biology", + "task_alias": "High Biology", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Biology", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_civics": { + "task": "arabicmmlu_high_civics", + "task_alias": "High Civics", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Civics", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_computer_science": { + "task": "arabicmmlu_high_computer_science", + "task_alias": "High Computer Science", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Computer Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_economics": { + "task": "arabicmmlu_high_economics", + "task_alias": "High Economics", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Economics", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_geography": { + "task": "arabicmmlu_high_geography", + "task_alias": "High Geography", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Geography", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_history": { + "task": "arabicmmlu_high_history", + "task_alias": "High History", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High History", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_islamic_studies": { + "task": "arabicmmlu_high_islamic_studies", + "task_alias": "High Islamic Studies", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Islamic Studies", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_philosophy": { + "task": "arabicmmlu_high_philosophy", + "task_alias": "High Philosophy", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Philosophy", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_high_physics": { + "task": "arabicmmlu_high_physics", + "task_alias": "High Physics", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "High Physics", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_islamic_studies": { + "task": "arabicmmlu_islamic_studies", + "task_alias": "Islamic Studies", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Islamic Studies", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_arabic_language": { + "task": "arabicmmlu_middle_arabic_language", + "task_alias": "Middle Arabic Language", + "tag": "arabicmmlu_language_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Arabic Language", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_civics": { + "task": "arabicmmlu_middle_civics", + "task_alias": "Middle Civics", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Civics", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_computer_science": { + "task": "arabicmmlu_middle_computer_science", + "task_alias": "Middle Computer Science", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Computer Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_economics": { + "task": "arabicmmlu_middle_economics", + "task_alias": "Middle Economics", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Economics", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_general_knowledge": { + "task": "arabicmmlu_middle_general_knowledge", + "task_alias": "Middle General Knowledge", + "tag": "arabicmmlu_other_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle General Knowledge", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_geography": { + "task": "arabicmmlu_middle_geography", + "task_alias": "Middle Geography", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Geography", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_history": { + "task": "arabicmmlu_middle_history", + "task_alias": "Middle History", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle History", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_islamic_studies": { + "task": "arabicmmlu_middle_islamic_studies", + "task_alias": "Middle Islamic Studies", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Islamic Studies", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_natural_science": { + "task": "arabicmmlu_middle_natural_science", + "task_alias": "Middle Natural Science", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Natural Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_middle_social_science": { + "task": "arabicmmlu_middle_social_science", + "task_alias": "Middle Social Science", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Middle Social Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_arabic_language": { + "task": "arabicmmlu_primary_arabic_language", + "task_alias": "Primary Arabic Language", + "tag": "arabicmmlu_language_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary Arabic Language", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_computer_science": { + "task": "arabicmmlu_primary_computer_science", + "task_alias": "Primary Computer Science", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary Computer Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_general_knowledge": { + "task": "arabicmmlu_primary_general_knowledge", + "task_alias": "Primary General Knowledge", + "tag": "arabicmmlu_other_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary General Knowledge", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_geography": { + "task": "arabicmmlu_primary_geography", + "task_alias": "Primary Geography", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary Geography", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_history": { + "task": "arabicmmlu_primary_history", + "task_alias": "Primary History", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary History", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_islamic_studies": { + "task": "arabicmmlu_primary_islamic_studies", + "task_alias": "Primary Islamic Studies", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary Islamic Studies", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_math": { + "task": "arabicmmlu_primary_math", + "task_alias": "Primary Math", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary Math", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_natural_science": { + "task": "arabicmmlu_primary_natural_science", + "task_alias": "Primary Natural Science", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary Natural Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_primary_social_science": { + "task": "arabicmmlu_primary_social_science", + "task_alias": "Primary Social Science", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Primary Social Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_prof_law": { + "task": "arabicmmlu_prof_law", + "task_alias": "Prof Law", + "tag": "arabicmmlu_humanities_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Prof Law", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_univ_accounting": { + "task": "arabicmmlu_univ_accounting", + "task_alias": "Univ Accounting", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Univ Accounting", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_univ_computer_science": { + "task": "arabicmmlu_univ_computer_science", + "task_alias": "Univ Computer Science", + "tag": "arabicmmlu_stem_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Univ Computer Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_univ_economics": { + "task": "arabicmmlu_univ_economics", + "task_alias": "Univ Economics", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Univ Economics", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_univ_management": { + "task": "arabicmmlu_univ_management", + "task_alias": "Univ Management", + "tag": "arabicmmlu_other_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Univ Management", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + }, + "arabicmmlu_univ_political_science": { + "task": "arabicmmlu_univ_political_science", + "task_alias": "Univ Political Science", + "tag": "arabicmmlu_social_science_tasks", + "dataset_path": "yazeed7/ArabicMMLU", + "dataset_name": "Univ Political Science", + "test_split": "test", + "fewshot_split": "dev", + "doc_to_text": "def doc_to_text(doc):\n \"\"\"\n Refactoring `prepare_data_en` to fit with the lm harness framework.\n https://github.com/mbzuai-nlp/ArabicMMLU/blob/main/util_prompt.py\n \"\"\"\n\n level = \"\" if not doc[\"Level\"] else \" for \" + level_en[doc[\"Level\"]]\n country = \"\" if not doc[\"Country\"] else \" in \" + doc[\"Country\"]\n main_meta_data = f\"{doc['Subject']} question{level}{country}\"\n\n question = (\n doc[\"Question\"]\n if doc[\"Context\"] == \"\"\n else f\"{doc['Context']}\\n\\n{doc['Question']}\"\n )\n\n options = []\n for i, opt in enumerate(\n [\"Option 1\", \"Option 2\", \"Option 3\", \"Option 4\", \"Option 5\"]\n ):\n if not doc[opt]:\n break\n options.append(f\"{alpa[i]} {doc[opt]}\")\n\n doc_text = PROMPT.format(main_meta_data, question, \"\\n\".join(options))\n\n return doc_text\n", + "doc_to_target": "Answer Key", + "doc_to_choice": "def doc_to_choice(doc):\n return [alpa[i][0] for i in range(5) if doc[f\"Option {i+1}\"]]\n", + "description": "", + "target_delimiter": " ", + "fewshot_delimiter": "\n\n", + "fewshot_config": { + "sampler": "first_n" + }, + "num_fewshot": 0, + "metric_list": [ + { + "metric": "acc", + "aggregation": "mean", + "higher_is_better": true + } + ], + "output_type": "multiple_choice", + "repeats": 1, + "should_decontaminate": false, + "metadata": { + "version": 0.0 + } + } + }, + "versions": { + "arabicmmlu": 0, + "arabicmmlu_arabic_language_(general)": 0.0, + "arabicmmlu_arabic_language_(grammar)": 0.0, + "arabicmmlu_driving_test": 0.0, + "arabicmmlu_general_knowledge": 0.0, + "arabicmmlu_high_arabic_language": 0.0, + "arabicmmlu_high_biology": 0.0, + "arabicmmlu_high_civics": 0.0, + "arabicmmlu_high_computer_science": 0.0, + "arabicmmlu_high_economics": 0.0, + "arabicmmlu_high_geography": 0.0, + "arabicmmlu_high_history": 0.0, + "arabicmmlu_high_islamic_studies": 0.0, + "arabicmmlu_high_philosophy": 0.0, + "arabicmmlu_high_physics": 0.0, + "arabicmmlu_humanities": 0, + "arabicmmlu_islamic_studies": 0.0, + "arabicmmlu_language": 0, + "arabicmmlu_middle_arabic_language": 0.0, + "arabicmmlu_middle_civics": 0.0, + "arabicmmlu_middle_computer_science": 0.0, + "arabicmmlu_middle_economics": 0.0, + "arabicmmlu_middle_general_knowledge": 0.0, + "arabicmmlu_middle_geography": 0.0, + "arabicmmlu_middle_history": 0.0, + "arabicmmlu_middle_islamic_studies": 0.0, + "arabicmmlu_middle_natural_science": 0.0, + "arabicmmlu_middle_social_science": 0.0, + "arabicmmlu_other": 0, + "arabicmmlu_primary_arabic_language": 0.0, + "arabicmmlu_primary_computer_science": 0.0, + "arabicmmlu_primary_general_knowledge": 0.0, + "arabicmmlu_primary_geography": 0.0, + "arabicmmlu_primary_history": 0.0, + "arabicmmlu_primary_islamic_studies": 0.0, + "arabicmmlu_primary_math": 0.0, + "arabicmmlu_primary_natural_science": 0.0, + "arabicmmlu_primary_social_science": 0.0, + "arabicmmlu_prof_law": 0.0, + "arabicmmlu_social_science": 0, + "arabicmmlu_stem": 0, + "arabicmmlu_univ_accounting": 0.0, + "arabicmmlu_univ_computer_science": 0.0, + "arabicmmlu_univ_economics": 0.0, + "arabicmmlu_univ_management": 0.0, + "arabicmmlu_univ_political_science": 0.0 + }, + "n-shot": { + "arabicmmlu_arabic_language_(general)": 0, + "arabicmmlu_arabic_language_(grammar)": 0, + "arabicmmlu_driving_test": 0, + "arabicmmlu_general_knowledge": 0, + "arabicmmlu_high_arabic_language": 0, + "arabicmmlu_high_biology": 0, + "arabicmmlu_high_civics": 0, + "arabicmmlu_high_computer_science": 0, + "arabicmmlu_high_economics": 0, + "arabicmmlu_high_geography": 0, + "arabicmmlu_high_history": 0, + "arabicmmlu_high_islamic_studies": 0, + "arabicmmlu_high_philosophy": 0, + "arabicmmlu_high_physics": 0, + "arabicmmlu_islamic_studies": 0, + "arabicmmlu_middle_arabic_language": 0, + "arabicmmlu_middle_civics": 0, + "arabicmmlu_middle_computer_science": 0, + "arabicmmlu_middle_economics": 0, + "arabicmmlu_middle_general_knowledge": 0, + "arabicmmlu_middle_geography": 0, + "arabicmmlu_middle_history": 0, + "arabicmmlu_middle_islamic_studies": 0, + "arabicmmlu_middle_natural_science": 0, + "arabicmmlu_middle_social_science": 0, + "arabicmmlu_primary_arabic_language": 0, + "arabicmmlu_primary_computer_science": 0, + "arabicmmlu_primary_general_knowledge": 0, + "arabicmmlu_primary_geography": 0, + "arabicmmlu_primary_history": 0, + "arabicmmlu_primary_islamic_studies": 0, + "arabicmmlu_primary_math": 0, + "arabicmmlu_primary_natural_science": 0, + "arabicmmlu_primary_social_science": 0, + "arabicmmlu_prof_law": 0, + "arabicmmlu_univ_accounting": 0, + "arabicmmlu_univ_computer_science": 0, + "arabicmmlu_univ_economics": 0, + "arabicmmlu_univ_management": 0, + "arabicmmlu_univ_political_science": 0 + }, + "higher_is_better": { + "arabicmmlu": { + "acc": true + }, + "arabicmmlu_arabic_language_(general)": { + "acc": true + }, + "arabicmmlu_arabic_language_(grammar)": { + "acc": true + }, + "arabicmmlu_driving_test": { + "acc": true + }, + "arabicmmlu_general_knowledge": { + "acc": true + }, + "arabicmmlu_high_arabic_language": { + "acc": true + }, + "arabicmmlu_high_biology": { + "acc": true + }, + "arabicmmlu_high_civics": { + "acc": true + }, + "arabicmmlu_high_computer_science": { + "acc": true + }, + "arabicmmlu_high_economics": { + "acc": true + }, + "arabicmmlu_high_geography": { + "acc": true + }, + "arabicmmlu_high_history": { + "acc": true + }, + "arabicmmlu_high_islamic_studies": { + "acc": true + }, + "arabicmmlu_high_philosophy": { + "acc": true + }, + "arabicmmlu_high_physics": { + "acc": true + }, + "arabicmmlu_humanities": { + "acc": true + }, + "arabicmmlu_islamic_studies": { + "acc": true + }, + "arabicmmlu_language": { + "acc": true + }, + "arabicmmlu_middle_arabic_language": { + "acc": true + }, + "arabicmmlu_middle_civics": { + "acc": true + }, + "arabicmmlu_middle_computer_science": { + "acc": true + }, + "arabicmmlu_middle_economics": { + "acc": true + }, + "arabicmmlu_middle_general_knowledge": { + "acc": true + }, + "arabicmmlu_middle_geography": { + "acc": true + }, + "arabicmmlu_middle_history": { + "acc": true + }, + "arabicmmlu_middle_islamic_studies": { + "acc": true + }, + "arabicmmlu_middle_natural_science": { + "acc": true + }, + "arabicmmlu_middle_social_science": { + "acc": true + }, + "arabicmmlu_other": { + "acc": true + }, + "arabicmmlu_primary_arabic_language": { + "acc": true + }, + "arabicmmlu_primary_computer_science": { + "acc": true + }, + "arabicmmlu_primary_general_knowledge": { + "acc": true + }, + "arabicmmlu_primary_geography": { + "acc": true + }, + "arabicmmlu_primary_history": { + "acc": true + }, + "arabicmmlu_primary_islamic_studies": { + "acc": true + }, + "arabicmmlu_primary_math": { + "acc": true + }, + "arabicmmlu_primary_natural_science": { + "acc": true + }, + "arabicmmlu_primary_social_science": { + "acc": true + }, + "arabicmmlu_prof_law": { + "acc": true + }, + "arabicmmlu_social_science": { + "acc": true + }, + "arabicmmlu_stem": { + "acc": true + }, + "arabicmmlu_univ_accounting": { + "acc": true + }, + "arabicmmlu_univ_computer_science": { + "acc": true + }, + "arabicmmlu_univ_economics": { + "acc": true + }, + "arabicmmlu_univ_management": { + "acc": true + }, + "arabicmmlu_univ_political_science": { + "acc": true + } + }, + "n-samples": { + "arabicmmlu_primary_general_knowledge": { + "original": 162, + "effective": 162 + }, + "arabicmmlu_general_knowledge": { + "original": 864, + "effective": 864 + }, + "arabicmmlu_middle_general_knowledge": { + "original": 172, + "effective": 172 + }, + "arabicmmlu_univ_management": { + "original": 75, + "effective": 75 + }, + "arabicmmlu_driving_test": { + "original": 1211, + "effective": 1211 + }, + "arabicmmlu_middle_social_science": { + "original": 241, + "effective": 241 + }, + "arabicmmlu_univ_economics": { + "original": 137, + "effective": 137 + }, + "arabicmmlu_univ_accounting": { + "original": 74, + "effective": 74 + }, + "arabicmmlu_high_civics": { + "original": 87, + "effective": 87 + }, + "arabicmmlu_high_economics": { + "original": 360, + "effective": 360 + }, + "arabicmmlu_middle_geography": { + "original": 272, + "effective": 272 + }, + "arabicmmlu_primary_geography": { + "original": 57, + "effective": 57 + }, + "arabicmmlu_middle_civics": { + "original": 236, + "effective": 236 + }, + "arabicmmlu_high_geography": { + "original": 1038, + "effective": 1038 + }, + "arabicmmlu_middle_economics": { + "original": 87, + "effective": 87 + }, + "arabicmmlu_univ_political_science": { + "original": 210, + "effective": 210 + }, + "arabicmmlu_primary_social_science": { + "original": 705, + "effective": 705 + }, + "arabicmmlu_prof_law": { + "original": 314, + "effective": 314 + }, + "arabicmmlu_middle_islamic_studies": { + "original": 238, + "effective": 238 + }, + "arabicmmlu_high_philosophy": { + "original": 39, + "effective": 39 + }, + "arabicmmlu_high_islamic_studies": { + "original": 334, + "effective": 334 + }, + "arabicmmlu_islamic_studies": { + "original": 639, + "effective": 639 + }, + "arabicmmlu_high_history": { + "original": 760, + "effective": 760 + }, + "arabicmmlu_primary_islamic_studies": { + "original": 999, + "effective": 999 + }, + "arabicmmlu_middle_history": { + "original": 203, + "effective": 203 + }, + "arabicmmlu_primary_history": { + "original": 102, + "effective": 102 + }, + "arabicmmlu_high_computer_science": { + "original": 261, + "effective": 261 + }, + "arabicmmlu_primary_math": { + "original": 409, + "effective": 409 + }, + "arabicmmlu_high_biology": { + "original": 1409, + "effective": 1409 + }, + "arabicmmlu_primary_computer_science": { + "original": 190, + "effective": 190 + }, + "arabicmmlu_middle_natural_science": { + "original": 242, + "effective": 242 + }, + "arabicmmlu_high_physics": { + "original": 255, + "effective": 255 + }, + "arabicmmlu_middle_computer_science": { + "original": 27, + "effective": 27 + }, + "arabicmmlu_univ_computer_science": { + "original": 64, + "effective": 64 + }, + "arabicmmlu_primary_natural_science": { + "original": 336, + "effective": 336 + }, + "arabicmmlu_high_arabic_language": { + "original": 390, + "effective": 390 + }, + "arabicmmlu_arabic_language_(grammar)": { + "original": 365, + "effective": 365 + }, + "arabicmmlu_middle_arabic_language": { + "original": 27, + "effective": 27 + }, + "arabicmmlu_arabic_language_(general)": { + "original": 612, + "effective": 612 + }, + "arabicmmlu_primary_arabic_language": { + "original": 252, + "effective": 252 + } + }, + "config": { + "model": "vllm", + "model_args": "pretrained=/tmp/7b-alpha-v1.27.2.25,tensor_parallel_size=1,data_parallel_size=2,gpu_memory_utilization=0.8", + "batch_size": 1, + "batch_sizes": [], + "device": null, + "use_cache": null, + "limit": null, + "bootstrap_iters": 100000, + "gen_kwargs": null, + "random_seed": 0, + "numpy_seed": 1234, + "torch_seed": 1234, + "fewshot_seed": 1234 + }, + "git_hash": "8e1bd48d", + "date": 1735662320.4500997, + "pretty_env_info": "PyTorch version: 2.4.0+cu121\nIs debug build: False\nCUDA used to build PyTorch: 12.1\nROCM used to build PyTorch: N/A\n\nOS: Ubuntu 22.04.3 LTS (x86_64)\nGCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0\nClang version: Could not collect\nCMake version: version 3.27.1\nLibc version: glibc-2.35\n\nPython version: 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0] (64-bit runtime)\nPython platform: Linux-5.15.0-1064-azure-x86_64-with-glibc2.35\nIs CUDA available: True\nCUDA runtime version: 12.2.128\nCUDA_MODULE_LOADING set to: LAZY\nGPU models and configuration: \nGPU 0: NVIDIA A100 80GB PCIe\nGPU 1: NVIDIA A100 80GB PCIe\n\nNvidia driver version: 535.161.08\ncuDNN version: Probably one of the following:\n/usr/lib/x86_64-linux-gnu/libcudnn.so.8.9.4\n/usr/lib/x86_64-linux-gnu/libcudnn_adv_infer.so.8.9.4\n/usr/lib/x86_64-linux-gnu/libcudnn_adv_train.so.8.9.4\n/usr/lib/x86_64-linux-gnu/libcudnn_cnn_infer.so.8.9.4\n/usr/lib/x86_64-linux-gnu/libcudnn_cnn_train.so.8.9.4\n/usr/lib/x86_64-linux-gnu/libcudnn_ops_infer.so.8.9.4\n/usr/lib/x86_64-linux-gnu/libcudnn_ops_train.so.8.9.4\nHIP runtime version: N/A\nMIOpen runtime version: N/A\nIs XNNPACK available: True\n\nCPU:\nArchitecture: x86_64\nCPU op-mode(s): 32-bit, 64-bit\nAddress sizes: 48 bits physical, 48 bits virtual\nByte Order: Little Endian\nCPU(s): 48\nOn-line CPU(s) list: 0-47\nVendor ID: AuthenticAMD\nModel name: AMD EPYC 7V13 64-Core Processor\nCPU family: 25\nModel: 1\nThread(s) per core: 1\nCore(s) per socket: 48\nSocket(s): 1\nStepping: 1\nBogoMIPS: 4890.87\nFlags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl tsc_reliable nonstop_tsc cpuid extd_apicid aperfmperf pni pclmulqdq ssse3 fma cx16 pcid sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand hypervisor lahf_lm cmp_legacy cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw topoext perfctr_core invpcid_single vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves clzero xsaveerptr rdpru arat umip vaes vpclmulqdq rdpid fsrm\nHypervisor vendor: Microsoft\nVirtualization type: full\nL1d cache: 1.5 MiB (48 instances)\nL1i cache: 1.5 MiB (48 instances)\nL2 cache: 24 MiB (48 instances)\nL3 cache: 192 MiB (6 instances)\nNUMA node(s): 2\nNUMA node0 CPU(s): 0-23\nNUMA node1 CPU(s): 24-47\nVulnerability Gather data sampling: Not affected\nVulnerability Itlb multihit: Not affected\nVulnerability L1tf: Not affected\nVulnerability Mds: Not affected\nVulnerability Meltdown: Not affected\nVulnerability Mmio stale data: Not affected\nVulnerability Retbleed: Not affected\nVulnerability Spec rstack overflow: Mitigation; safe RET, no microcode\nVulnerability Spec store bypass: Vulnerable\nVulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization\nVulnerability Spectre v2: Mitigation; Retpolines; STIBP disabled; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected\nVulnerability Srbds: Not affected\nVulnerability Tsx async abort: Not affected\n\nVersions of relevant libraries:\n[pip3] numpy==1.26.4\n[pip3] onnx==1.14.0\n[pip3] pytorch-lightning==2.0.7\n[pip3] pytorch-quantization==2.1.2\n[pip3] torch==2.4.0\n[pip3] torch-tensorrt==2.0.0.dev0\n[pip3] torchaudio==2.1.0\n[pip3] torchdata==0.7.0a0\n[pip3] torchmetrics==1.2.0\n[pip3] torchvision==0.19.0\n[pip3] triton==3.0.0\n[conda] Could not collect", + "transformers_version": "4.47.1", + "upper_git_hash": null, + "tokenizer_pad_token": [ + "", + "0" + ], + "tokenizer_eos_token": [ + "", + "2" + ], + "tokenizer_bos_token": [ + "", + "1" + ], + "eot_token_id": 2, + "max_length": 4096, + "task_hashes": { + "arabicmmlu_primary_general_knowledge": "9c41f9b2409e40ac46be285d8ef0c425c69f2e89f389af149388ed3317803f47", + "arabicmmlu_general_knowledge": "d0d398d26921bf02c874c7f6261b3b35569d2e5d4f5ff0b57c3849702ac76c7d", + "arabicmmlu_middle_general_knowledge": "01dc69e7e4349d3ad2d4c3a1aa9c3223aa6b80b49eb927328995d78a7119d12e", + "arabicmmlu_univ_management": "a75412840fc2690239048b87ff63c88576d098043214e33c0f893ae262adf558", + "arabicmmlu_driving_test": "1294a352f9996956b5eb556dfb4ad8da6c107cf83d78057e03423a1d263271eb", + "arabicmmlu_middle_social_science": "aaa200ab5bef99e627e5cc2339616fe893324ba9f0e6bc21b1cbf50fb12f87a4", + "arabicmmlu_univ_economics": "ec1e184a96e1c5fb9ebcf75c7a681987e10269f310970712fa7e08cf08aedf9c", + "arabicmmlu_univ_accounting": "e86c0c589105cd0a8799c9f9ed5d3be8fd66a372b0c276d841224253ac26caf3", + "arabicmmlu_high_civics": "1782368ed0854ebb92d306d63b5309220d9dbc812e759134bdb319a4798a9f4a", + "arabicmmlu_high_economics": "98ec2aac658625844ae7905b5bbb20e9b1d008e80237fac4562d269c98d95036", + "arabicmmlu_middle_geography": "11b273709d3739cd0ca0112960b7f80126185838d2573abf434f4d13b1b58a41", + "arabicmmlu_primary_geography": "280a1771b756a73d2e6ded00eecadbac20e4ee1ef00949a3b0825e9d997c6125", + "arabicmmlu_middle_civics": "ddbc97ff3f96ceaff0e296b6c9bf792f50d50f076200ca9a60bf72137508246d", + "arabicmmlu_high_geography": "faf4ba7fc6c07d9d395ab8b3cf1d3f62d2aa51297d1de2417503d99725ee5968", + "arabicmmlu_middle_economics": "411a71e9a0975e178836323da11af60b68483e80e6e50c16e8ab5a4399b15cf6", + "arabicmmlu_univ_political_science": "1b4e81c09070ed52587d966e92a753718fd6afc4f22b885a75aeca950f7bbc44", + "arabicmmlu_primary_social_science": "14b9797e030d4915891382e67f531aff407f495a0c95de390cb140415da4853e", + "arabicmmlu_prof_law": "929be8388dbe8a64e52db14f2d17ab627b51fa59718b97bab57d7f885ae22745", + "arabicmmlu_middle_islamic_studies": "212f989ad1b21aa4d465b9eac1f49cbc7885f57130768926cc6b44299bab862b", + "arabicmmlu_high_philosophy": "7918cb8aff5e2ce06d60f7b8a476db496f12f1c528a5c76dab4e1a7a3802615f", + "arabicmmlu_high_islamic_studies": "36c0092e41cc9b74cf95e7580a22cd3bc6c1c8be1b583aeef612303a644ee5d1", + "arabicmmlu_islamic_studies": "61441e32632d46ba8de49eb0db6c9424402d26c7cfd21cf80cad845f78162d25", + "arabicmmlu_high_history": "db21ec3b92313a8ff84eea1ef253bd9fd311b799b7255530752c9d9d42582e31", + "arabicmmlu_primary_islamic_studies": "948fda0d0bc5d6b7f3d4778361317c5f1ccd749e82071cec7710ebe034f8e5cf", + "arabicmmlu_middle_history": "06d1eee1e75a711e0f6e4b6209b1ddf2b7b9ac8fd4e9e19c83bc260664e9da92", + "arabicmmlu_primary_history": "236ef1dc7fe81ba7e3abf7f4c0f706e5cf1932692f6bb670df7fcdd8118843ee", + "arabicmmlu_high_computer_science": "b94390a6fd058297d59d43575ce189c833d75fd636894320989d8628b074f002", + "arabicmmlu_primary_math": "7fbd73f73bc85611f0495ed87530d6512d9da9e0c92fe25553a591b91ef4e79d", + "arabicmmlu_high_biology": "daeac852f0eb44834936f0a04bc71521d2b9d939d47e7976b80f1e576b7688c1", + "arabicmmlu_primary_computer_science": "bb40dbb3bf51122ea2a0cc30848e010b71de881a8b7a6b5f11e97c36867431e6", + "arabicmmlu_middle_natural_science": "5d3ab2bf4ca8633ecf28783ae2d05d0025d3af21add23eadd96cea54c63427cf", + "arabicmmlu_high_physics": "defccd1d721b1ba615956f253ad5f61f383b5f8a9d2aae786b58bbd212f87ec1", + "arabicmmlu_middle_computer_science": "6d88646a6979333723a7697392ef4bf8d9440001ebe886ca85f5461f3a510048", + "arabicmmlu_univ_computer_science": "1e38d7bfc8a18b04cc9e57e3ae4e3c11f4d4fc6f07321feba0d36a3122923d0b", + "arabicmmlu_primary_natural_science": "fac384e5d9b22d1c20239d6d2563d9f0a79fb48cf615204fcf229fc37c76a008", + "arabicmmlu_high_arabic_language": "f4771e89a45e43ae733dcfda251963f5de5383f783d5f534e4ce1999a67b6116", + "arabicmmlu_arabic_language_(grammar)": "17e3b209cf3c2d60d47089cdcfdd29f18f8af73b5b9ef05fe6207dfaa0d4c41b", + "arabicmmlu_middle_arabic_language": "3332b66219055daebf1b147ad8f648a3edcc672ef99feb2ded597ae8740a995c", + "arabicmmlu_arabic_language_(general)": "baa8d90299504f0ee7dd6b57071cf0502218545f926847cd2f30b92be8aeed8b", + "arabicmmlu_primary_arabic_language": "70a513c8c604cd2edb7ab15dea6e21908f1a4136dbd98e3a1294a7111dfa4228" + }, + "model_source": "vllm", + "model_name": "/tmp/7b-alpha-v1.27.2.25", + "model_name_sanitized": "__tmp__7b-alpha-v1.27.2.25", + "system_instruction": null, + "system_instruction_sha": null, + "fewshot_as_multiturn": false, + "chat_template": null, + "chat_template_sha": null, + "start_time": 2980.642859002, + "end_time": 3340.273846829, + "total_evaluation_time_seconds": "359.6309878269999" +} \ No newline at end of file