2051 lines
99 KiB
JSON
2051 lines
99 KiB
JSON
|
{
|
||
|
|
"results": {
|
||
|
|
"arabicmmlu": {
|
||
|
|
"acc,none": 0.564303009339329,
|
||
|
|
"acc_stderr,none": 0.0040196752630034735,
|
||
|
|
"alias": "arabicmmlu"
|
||
|
|
},
|
||
|
|
"arabicmmlu_humanities": {
|
||
|
|
"acc,none": 0.5587100330760749,
|
||
|
|
"acc_stderr,none": 0.007915141829477251,
|
||
|
|
"alias": " - Humanities"
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_history": {
|
||
|
|
"alias": " - High History",
|
||
|
|
"acc,none": 0.4276315789473684,
|
||
|
|
"acc_stderr,none": 0.01795774617649965
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_islamic_studies": {
|
||
|
|
"alias": " - High Islamic Studies",
|
||
|
|
"acc,none": 0.6766467065868264,
|
||
|
|
"acc_stderr,none": 0.02563288645517917
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_philosophy": {
|
||
|
|
"alias": " - High Philosophy",
|
||
|
|
"acc,none": 0.5641025641025641,
|
||
|
|
"acc_stderr,none": 0.08044135838502685
|
||
|
|
},
|
||
|
|
"arabicmmlu_islamic_studies": {
|
||
|
|
"alias": " - Islamic Studies",
|
||
|
|
"acc,none": 0.38341158059467917,
|
||
|
|
"acc_stderr,none": 0.01924952226173331
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_history": {
|
||
|
|
"alias": " - Middle History",
|
||
|
|
"acc,none": 0.5960591133004927,
|
||
|
|
"acc_stderr,none": 0.03452453903822032
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_islamic_studies": {
|
||
|
|
"alias": " - Middle Islamic Studies",
|
||
|
|
"acc,none": 0.6764705882352942,
|
||
|
|
"acc_stderr,none": 0.030388353551886797
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_history": {
|
||
|
|
"alias": " - Primary History",
|
||
|
|
"acc,none": 0.5392156862745098,
|
||
|
|
"acc_stderr,none": 0.049598599663841815
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_islamic_studies": {
|
||
|
|
"alias": " - Primary Islamic Studies",
|
||
|
|
"acc,none": 0.7267267267267268,
|
||
|
|
"acc_stderr,none": 0.014106487065973238
|
||
|
|
},
|
||
|
|
"arabicmmlu_prof_law": {
|
||
|
|
"alias": " - Prof Law",
|
||
|
|
"acc,none": 0.46496815286624205,
|
||
|
|
"acc_stderr,none": 0.02819221844954206
|
||
|
|
},
|
||
|
|
"arabicmmlu_language": {
|
||
|
|
"acc,none": 0.56318347509113,
|
||
|
|
"acc_stderr,none": 0.011882048451256877,
|
||
|
|
"alias": " - Language"
|
||
|
|
},
|
||
|
|
"arabicmmlu_arabic_language_(general)": {
|
||
|
|
"alias": " - Arabic Language (General)",
|
||
|
|
"acc,none": 0.6683006535947712,
|
||
|
|
"acc_stderr,none": 0.019047485239360375
|
||
|
|
},
|
||
|
|
"arabicmmlu_arabic_language_(grammar)": {
|
||
|
|
"alias": " - Arabic Language (Grammar)",
|
||
|
|
"acc,none": 0.5698630136986301,
|
||
|
|
"acc_stderr,none": 0.02595003437064698
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_arabic_language": {
|
||
|
|
"alias": " - High Arabic Language",
|
||
|
|
"acc,none": 0.3641025641025641,
|
||
|
|
"acc_stderr,none": 0.02439667298509477
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_arabic_language": {
|
||
|
|
"alias": " - Middle Arabic Language",
|
||
|
|
"acc,none": 0.7777777777777778,
|
||
|
|
"acc_stderr,none": 0.08153326507837146
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_arabic_language": {
|
||
|
|
"alias": " - Primary Arabic Language",
|
||
|
|
"acc,none": 0.5833333333333334,
|
||
|
|
"acc_stderr,none": 0.031118303728104594
|
||
|
|
},
|
||
|
|
"arabicmmlu_other": {
|
||
|
|
"acc,none": 0.6272141706924316,
|
||
|
|
"acc_stderr,none": 0.009640611430777322,
|
||
|
|
"alias": " - Other"
|
||
|
|
},
|
||
|
|
"arabicmmlu_driving_test": {
|
||
|
|
"alias": " - Driving Test",
|
||
|
|
"acc,none": 0.6672171758876961,
|
||
|
|
"acc_stderr,none": 0.013546321390449041
|
||
|
|
},
|
||
|
|
"arabicmmlu_general_knowledge": {
|
||
|
|
"alias": " - General Knowledge",
|
||
|
|
"acc,none": 0.5474537037037037,
|
||
|
|
"acc_stderr,none": 0.016943370542362845
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_general_knowledge": {
|
||
|
|
"alias": " - Middle General Knowledge",
|
||
|
|
"acc,none": 0.6686046511627907,
|
||
|
|
"acc_stderr,none": 0.035996464381795934
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_general_knowledge": {
|
||
|
|
"alias": " - Primary General Knowledge",
|
||
|
|
"acc,none": 0.6851851851851852,
|
||
|
|
"acc_stderr,none": 0.036603163762720714
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_management": {
|
||
|
|
"alias": " - Univ Management",
|
||
|
|
"acc,none": 0.68,
|
||
|
|
"acc_stderr,none": 0.05422675115236518
|
||
|
|
},
|
||
|
|
"arabicmmlu_social_science": {
|
||
|
|
"acc,none": 0.5547945205479452,
|
||
|
|
"acc_stderr,none": 0.008278003487917672,
|
||
|
|
"alias": " - Social Science"
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_civics": {
|
||
|
|
"alias": " - High Civics",
|
||
|
|
"acc,none": 0.4367816091954023,
|
||
|
|
"acc_stderr,none": 0.05348368965287097
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_economics": {
|
||
|
|
"alias": " - High Economics",
|
||
|
|
"acc,none": 0.575,
|
||
|
|
"acc_stderr,none": 0.026090425569673736
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_geography": {
|
||
|
|
"alias": " - High Geography",
|
||
|
|
"acc,none": 0.47398843930635837,
|
||
|
|
"acc_stderr,none": 0.015505727274549675
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_civics": {
|
||
|
|
"alias": " - Middle Civics",
|
||
|
|
"acc,none": 0.4872881355932203,
|
||
|
|
"acc_stderr,none": 0.03260586088180842
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_economics": {
|
||
|
|
"alias": " - Middle Economics",
|
||
|
|
"acc,none": 0.6666666666666666,
|
||
|
|
"acc_stderr,none": 0.05083285677753486
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_geography": {
|
||
|
|
"alias": " - Middle Geography",
|
||
|
|
"acc,none": 0.5845588235294118,
|
||
|
|
"acc_stderr,none": 0.029935342707877746
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_social_science": {
|
||
|
|
"alias": " - Middle Social Science",
|
||
|
|
"acc,none": 0.5228215767634855,
|
||
|
|
"acc_stderr,none": 0.03224122462224077
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_geography": {
|
||
|
|
"alias": " - Primary Geography",
|
||
|
|
"acc,none": 0.5789473684210527,
|
||
|
|
"acc_stderr,none": 0.06597717584505354
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_social_science": {
|
||
|
|
"alias": " - Primary Social Science",
|
||
|
|
"acc,none": 0.7021276595744681,
|
||
|
|
"acc_stderr,none": 0.017236012495765663
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_accounting": {
|
||
|
|
"alias": " - Univ Accounting",
|
||
|
|
"acc,none": 0.5675675675675675,
|
||
|
|
"acc_stderr,none": 0.057983774751431016
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_economics": {
|
||
|
|
"alias": " - Univ Economics",
|
||
|
|
"acc,none": 0.5547445255474452,
|
||
|
|
"acc_stderr,none": 0.04261688398864188
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_political_science": {
|
||
|
|
"alias": " - Univ Political Science",
|
||
|
|
"acc,none": 0.49047619047619045,
|
||
|
|
"acc_stderr,none": 0.034579448570031264
|
||
|
|
},
|
||
|
|
"arabicmmlu_stem": {
|
||
|
|
"acc,none": 0.5327278421547135,
|
||
|
|
"acc_stderr,none": 0.00860088193534487,
|
||
|
|
"alias": " - STEM"
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_biology": {
|
||
|
|
"alias": " - High Biology",
|
||
|
|
"acc,none": 0.43293115684882894,
|
||
|
|
"acc_stderr,none": 0.013204622401057848
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_computer_science": {
|
||
|
|
"alias": " - High Computer Science",
|
||
|
|
"acc,none": 0.5708812260536399,
|
||
|
|
"acc_stderr,none": 0.03069551782571805
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_physics": {
|
||
|
|
"alias": " - High Physics",
|
||
|
|
"acc,none": 0.43529411764705883,
|
||
|
|
"acc_stderr,none": 0.031108974626602753
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_computer_science": {
|
||
|
|
"alias": " - Middle Computer Science",
|
||
|
|
"acc,none": 0.7407407407407407,
|
||
|
|
"acc_stderr,none": 0.08594360757264022
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_natural_science": {
|
||
|
|
"alias": " - Middle Natural Science",
|
||
|
|
"acc,none": 0.6818181818181818,
|
||
|
|
"acc_stderr,none": 0.03000291471043612
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_computer_science": {
|
||
|
|
"alias": " - Primary Computer Science",
|
||
|
|
"acc,none": 0.6894736842105263,
|
||
|
|
"acc_stderr,none": 0.03365713545671698
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_math": {
|
||
|
|
"alias": " - Primary Math",
|
||
|
|
"acc,none": 0.5599022004889975,
|
||
|
|
"acc_stderr,none": 0.024575400500226115
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_natural_science": {
|
||
|
|
"alias": " - Primary Natural Science",
|
||
|
|
"acc,none": 0.7380952380952381,
|
||
|
|
"acc_stderr,none": 0.02402179716619147
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_computer_science": {
|
||
|
|
"alias": " - Univ Computer Science",
|
||
|
|
"acc,none": 0.59375,
|
||
|
|
"acc_stderr,none": 0.061876853828249374
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"groups": {
|
||
|
|
"arabicmmlu": {
|
||
|
|
"acc,none": 0.564303009339329,
|
||
|
|
"acc_stderr,none": 0.0040196752630034735,
|
||
|
|
"alias": "arabicmmlu"
|
||
|
|
},
|
||
|
|
"arabicmmlu_humanities": {
|
||
|
|
"acc,none": 0.5587100330760749,
|
||
|
|
"acc_stderr,none": 0.007915141829477251,
|
||
|
|
"alias": " - Humanities"
|
||
|
|
},
|
||
|
|
"arabicmmlu_language": {
|
||
|
|
"acc,none": 0.56318347509113,
|
||
|
|
"acc_stderr,none": 0.011882048451256877,
|
||
|
|
"alias": " - Language"
|
||
|
|
},
|
||
|
|
"arabicmmlu_other": {
|
||
|
|
"acc,none": 0.6272141706924316,
|
||
|
|
"acc_stderr,none": 0.009640611430777322,
|
||
|
|
"alias": " - Other"
|
||
|
|
},
|
||
|
|
"arabicmmlu_social_science": {
|
||
|
|
"acc,none": 0.5547945205479452,
|
||
|
|
"acc_stderr,none": 0.008278003487917672,
|
||
|
|
"alias": " - Social Science"
|
||
|
|
},
|
||
|
|
"arabicmmlu_stem": {
|
||
|
|
"acc,none": 0.5327278421547135,
|
||
|
|
"acc_stderr,none": 0.00860088193534487,
|
||
|
|
"alias": " - STEM"
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"group_subtasks": {
|
||
|
|
"arabicmmlu_language": [
|
||
|
|
"arabicmmlu_arabic_language_(general)",
|
||
|
|
"arabicmmlu_middle_arabic_language",
|
||
|
|
"arabicmmlu_primary_arabic_language",
|
||
|
|
"arabicmmlu_high_arabic_language",
|
||
|
|
"arabicmmlu_arabic_language_(grammar)"
|
||
|
|
],
|
||
|
|
"arabicmmlu_stem": [
|
||
|
|
"arabicmmlu_primary_math",
|
||
|
|
"arabicmmlu_primary_natural_science",
|
||
|
|
"arabicmmlu_middle_computer_science",
|
||
|
|
"arabicmmlu_high_physics",
|
||
|
|
"arabicmmlu_high_computer_science",
|
||
|
|
"arabicmmlu_high_biology",
|
||
|
|
"arabicmmlu_middle_natural_science",
|
||
|
|
"arabicmmlu_primary_computer_science",
|
||
|
|
"arabicmmlu_univ_computer_science"
|
||
|
|
],
|
||
|
|
"arabicmmlu_humanities": [
|
||
|
|
"arabicmmlu_high_islamic_studies",
|
||
|
|
"arabicmmlu_primary_islamic_studies",
|
||
|
|
"arabicmmlu_high_history",
|
||
|
|
"arabicmmlu_middle_islamic_studies",
|
||
|
|
"arabicmmlu_high_philosophy",
|
||
|
|
"arabicmmlu_middle_history",
|
||
|
|
"arabicmmlu_primary_history",
|
||
|
|
"arabicmmlu_islamic_studies",
|
||
|
|
"arabicmmlu_prof_law"
|
||
|
|
],
|
||
|
|
"arabicmmlu_social_science": [
|
||
|
|
"arabicmmlu_high_economics",
|
||
|
|
"arabicmmlu_high_civics",
|
||
|
|
"arabicmmlu_univ_accounting",
|
||
|
|
"arabicmmlu_middle_geography",
|
||
|
|
"arabicmmlu_primary_social_science",
|
||
|
|
"arabicmmlu_high_geography",
|
||
|
|
"arabicmmlu_middle_economics",
|
||
|
|
"arabicmmlu_univ_political_science",
|
||
|
|
"arabicmmlu_middle_social_science",
|
||
|
|
"arabicmmlu_univ_economics",
|
||
|
|
"arabicmmlu_primary_geography",
|
||
|
|
"arabicmmlu_middle_civics"
|
||
|
|
],
|
||
|
|
"arabicmmlu_other": [
|
||
|
|
"arabicmmlu_primary_general_knowledge",
|
||
|
|
"arabicmmlu_driving_test",
|
||
|
|
"arabicmmlu_univ_management",
|
||
|
|
"arabicmmlu_middle_general_knowledge",
|
||
|
|
"arabicmmlu_general_knowledge"
|
||
|
|
],
|
||
|
|
"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_driving_test": {
|
||
|
|
"original": 1211,
|
||
|
|
"effective": 1211
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_management": {
|
||
|
|
"original": 75,
|
||
|
|
"effective": 75
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_general_knowledge": {
|
||
|
|
"original": 172,
|
||
|
|
"effective": 172
|
||
|
|
},
|
||
|
|
"arabicmmlu_general_knowledge": {
|
||
|
|
"original": 864,
|
||
|
|
"effective": 864
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_economics": {
|
||
|
|
"original": 360,
|
||
|
|
"effective": 360
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_civics": {
|
||
|
|
"original": 87,
|
||
|
|
"effective": 87
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_accounting": {
|
||
|
|
"original": 74,
|
||
|
|
"effective": 74
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_geography": {
|
||
|
|
"original": 272,
|
||
|
|
"effective": 272
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_social_science": {
|
||
|
|
"original": 705,
|
||
|
|
"effective": 705
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_geography": {
|
||
|
|
"original": 1038,
|
||
|
|
"effective": 1038
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_economics": {
|
||
|
|
"original": 87,
|
||
|
|
"effective": 87
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_political_science": {
|
||
|
|
"original": 210,
|
||
|
|
"effective": 210
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_social_science": {
|
||
|
|
"original": 241,
|
||
|
|
"effective": 241
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_economics": {
|
||
|
|
"original": 137,
|
||
|
|
"effective": 137
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_geography": {
|
||
|
|
"original": 57,
|
||
|
|
"effective": 57
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_civics": {
|
||
|
|
"original": 236,
|
||
|
|
"effective": 236
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_islamic_studies": {
|
||
|
|
"original": 334,
|
||
|
|
"effective": 334
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_islamic_studies": {
|
||
|
|
"original": 999,
|
||
|
|
"effective": 999
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_history": {
|
||
|
|
"original": 760,
|
||
|
|
"effective": 760
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_islamic_studies": {
|
||
|
|
"original": 238,
|
||
|
|
"effective": 238
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_philosophy": {
|
||
|
|
"original": 39,
|
||
|
|
"effective": 39
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_history": {
|
||
|
|
"original": 203,
|
||
|
|
"effective": 203
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_history": {
|
||
|
|
"original": 102,
|
||
|
|
"effective": 102
|
||
|
|
},
|
||
|
|
"arabicmmlu_islamic_studies": {
|
||
|
|
"original": 639,
|
||
|
|
"effective": 639
|
||
|
|
},
|
||
|
|
"arabicmmlu_prof_law": {
|
||
|
|
"original": 314,
|
||
|
|
"effective": 314
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_math": {
|
||
|
|
"original": 409,
|
||
|
|
"effective": 409
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_natural_science": {
|
||
|
|
"original": 336,
|
||
|
|
"effective": 336
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_computer_science": {
|
||
|
|
"original": 27,
|
||
|
|
"effective": 27
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_physics": {
|
||
|
|
"original": 255,
|
||
|
|
"effective": 255
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_computer_science": {
|
||
|
|
"original": 261,
|
||
|
|
"effective": 261
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_biology": {
|
||
|
|
"original": 1409,
|
||
|
|
"effective": 1409
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_natural_science": {
|
||
|
|
"original": 242,
|
||
|
|
"effective": 242
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_computer_science": {
|
||
|
|
"original": 190,
|
||
|
|
"effective": 190
|
||
|
|
},
|
||
|
|
"arabicmmlu_univ_computer_science": {
|
||
|
|
"original": 64,
|
||
|
|
"effective": 64
|
||
|
|
},
|
||
|
|
"arabicmmlu_arabic_language_(general)": {
|
||
|
|
"original": 612,
|
||
|
|
"effective": 612
|
||
|
|
},
|
||
|
|
"arabicmmlu_middle_arabic_language": {
|
||
|
|
"original": 27,
|
||
|
|
"effective": 27
|
||
|
|
},
|
||
|
|
"arabicmmlu_primary_arabic_language": {
|
||
|
|
"original": 252,
|
||
|
|
"effective": 252
|
||
|
|
},
|
||
|
|
"arabicmmlu_high_arabic_language": {
|
||
|
|
"original": 390,
|
||
|
|
"effective": 390
|
||
|
|
},
|
||
|
|
"arabicmmlu_arabic_language_(grammar)": {
|
||
|
|
"original": 365,
|
||
|
|
"effective": 365
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"config": {
|
||
|
|
"model": "hf",
|
||
|
|
"model_args": "pretrained=meta-llama/Meta-Llama-3.1-8B-Instruct,trust_remote_code=True,cache_dir=/tmp,parallelize=False",
|
||
|
|
"model_num_parameters": 8030261248,
|
||
|
|
"model_dtype": "torch.bfloat16",
|
||
|
|
"model_revision": "main",
|
||
|
|
"model_sha": "0e9e39f249a16976918f6564b8830bc894c89659",
|
||
|
|
"batch_size": "auto",
|
||
|
|
"batch_sizes": [
|
||
|
|
32
|
||
|
|
],
|
||
|
|
"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": "788a3672",
|
||
|
|
"date": 1737778654.0503197,
|
||
|
|
"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.
|
||
|
|
"transformers_version": "4.48.1",
|
||
|
|
"upper_git_hash": "086919bd66f4e15fdcd4b792a7b27a698c1ba091",
|
||
|
|
"tokenizer_pad_token": [
|
||
|
|
"<|eot_id|>",
|
||
|
|
"128009"
|
||
|
|
],
|
||
|
|
"tokenizer_eos_token": [
|
||
|
|
"<|eot_id|>",
|
||
|
|
"128009"
|
||
|
|
],
|
||
|
|
"tokenizer_bos_token": [
|
||
|
|
"<|begin_of_text|>",
|
||
|
|
"128000"
|
||
|
|
],
|
||
|
|
"eot_token_id": 128009,
|
||
|
|
"max_length": 131072,
|
||
|
|
"task_hashes": {},
|
||
|
|
"model_source": "hf",
|
||
|
|
"model_name": "meta-llama/Meta-Llama-3.1-8B-Instruct",
|
||
|
|
"model_name_sanitized": "meta-llama__Meta-Llama-3.1-8B-Instruct",
|
||
|
|
"system_instruction": null,
|
||
|
|
"system_instruction_sha": null,
|
||
|
|
"fewshot_as_multiturn": false,
|
||
|
|
"chat_template": null,
|
||
|
|
"chat_template_sha": null,
|
||
|
|
"start_time": 25735.027525946,
|
||
|
|
"end_time": 25948.04309341,
|
||
|
|
"total_evaluation_time_seconds": "213.01556746400092"
|
||
|
|
}
|