metadata

license: other
license_name: creative-commons-by-nc
task_categories:
  - question-answering
language:
  - zh
tags:
  - traditional chinese
  - finance
  - medical
  - taiwan
  - benchmark
  - zh-tw
  - zh-hant
pretty_name: tmmlu++
size_categories:
  - 100K<n<1M
configs:
  - config_name: engineering_math
    datafiles:
      - split: train - "data/engineering_math_dev.csv"
      - split: validation - "data/engineering_math_val.csv"
      - split: test - "data/engineering_math_test.csv"
  - config_name: dentistry
    datafiles:
      - split: train - "data/dentistry_dev.csv"
      - split: validation - "data/dentistry_val.csv"
      - split: test - "data/dentistry_test.csv"
  - config_name: traditional_chinese_medicine_clinical_medicine
    datafiles:
      - split: train - "data/traditional_chinese_medicine_clinical_medicine_dev.csv"
      - split: >-
          validation -
          "data/traditional_chinese_medicine_clinical_medicine_val.csv"
      - split: test - "data/traditional_chinese_medicine_clinical_medicine_test.csv"
  - config_name: clinical_psychology
    datafiles:
      - split: train - "data/clinical_psychology_dev.csv"
      - split: validation - "data/clinical_psychology_val.csv"
      - split: test - "data/clinical_psychology_test.csv"
  - config_name: technical
    datafiles:
      - split: train - "data/technical_dev.csv"
      - split: validation - "data/technical_val.csv"
      - split: test - "data/technical_test.csv"
  - config_name: culinary_skills
    datafiles:
      - split: train - "data/culinary_skills_dev.csv"
      - split: validation - "data/culinary_skills_val.csv"
      - split: test - "data/culinary_skills_test.csv"
  - config_name: mechanical
    datafiles:
      - split: train - "data/mechanical_dev.csv"
      - split: validation - "data/mechanical_val.csv"
      - split: test - "data/mechanical_test.csv"
  - config_name: logic_reasoning
    datafiles:
      - split: train - "data/logic_reasoning_dev.csv"
      - split: validation - "data/logic_reasoning_val.csv"
      - split: test - "data/logic_reasoning_test.csv"
  - config_name: real_estate
    datafiles:
      - split: train - "data/real_estate_dev.csv"
      - split: validation - "data/real_estate_val.csv"
      - split: test - "data/real_estate_test.csv"
  - config_name: general_principles_of_law
    datafiles:
      - split: train - "data/general_principles_of_law_dev.csv"
      - split: validation - "data/general_principles_of_law_val.csv"
      - split: test - "data/general_principles_of_law_test.csv"
  - config_name: finance_banking
    datafiles:
      - split: train - "data/finance_banking_dev.csv"
      - split: validation - "data/finance_banking_val.csv"
      - split: test - "data/finance_banking_test.csv"
  - config_name: anti_money_laundering
    datafiles:
      - split: train - "data/anti_money_laundering_dev.csv"
      - split: validation - "data/anti_money_laundering_val.csv"
      - split: test - "data/anti_money_laundering_test.csv"
  - config_name: ttqav2
    datafiles:
      - split: train - "data/ttqav2_dev.csv"
      - split: validation - "data/ttqav2_val.csv"
      - split: test - "data/ttqav2_test.csv"
  - config_name: marketing_management
    datafiles:
      - split: train - "data/marketing_management_dev.csv"
      - split: validation - "data/marketing_management_val.csv"
      - split: test - "data/marketing_management_test.csv"
  - config_name: business_management
    datafiles:
      - split: train - "data/business_management_dev.csv"
      - split: validation - "data/business_management_val.csv"
      - split: test - "data/business_management_test.csv"
  - config_name: organic_chemistry
    datafiles:
      - split: train - "data/organic_chemistry_dev.csv"
      - split: validation - "data/organic_chemistry_val.csv"
      - split: test - "data/organic_chemistry_test.csv"
  - config_name: advance_chemistry
    datafiles:
      - split: train - "data/advance_chemistry_dev.csv"
      - split: validation - "data/advance_chemistry_val.csv"
      - split: test - "data/advance_chemistry_test.csv"
  - config_name: physics
    datafiles:
      - split: train - "data/physics_dev.csv"
      - split: validation - "data/physics_val.csv"
      - split: test - "data/physics_test.csv"
  - config_name: secondary_physics
    datafiles:
      - split: train - "data/secondary_physics_dev.csv"
      - split: validation - "data/secondary_physics_val.csv"
      - split: test - "data/secondary_physics_test.csv"
  - config_name: human_behavior
    datafiles:
      - split: train - "data/human_behavior_dev.csv"
      - split: validation - "data/human_behavior_val.csv"
      - split: test - "data/human_behavior_test.csv"
  - config_name: national_protection
    datafiles:
      - split: train - "data/national_protection_dev.csv"
      - split: validation - "data/national_protection_val.csv"
      - split: test - "data/national_protection_test.csv"
  - config_name: jce_humanities
    datafiles:
      - split: train - "data/jce_humanities_dev.csv"
      - split: validation - "data/jce_humanities_val.csv"
      - split: test - "data/jce_humanities_test.csv"
  - config_name: politic_science
    datafiles:
      - split: train - "data/politic_science_dev.csv"
      - split: validation - "data/politic_science_val.csv"
      - split: test - "data/politic_science_test.csv"
  - config_name: agriculture
    datafiles:
      - split: train - "data/agriculture_dev.csv"
      - split: validation - "data/agriculture_val.csv"
      - split: test - "data/agriculture_test.csv"
  - config_name: official_document_management
    datafiles:
      - split: train - "data/official_document_management_dev.csv"
      - split: validation - "data/official_document_management_val.csv"
      - split: test - "data/official_document_management_test.csv"
  - config_name: financial_analysis
    datafiles:
      - split: train - "data/financial_analysis_dev.csv"
      - split: validation - "data/financial_analysis_val.csv"
      - split: test - "data/financial_analysis_test.csv"
  - config_name: pharmacy
    datafiles:
      - split: train - "data/pharmacy_dev.csv"
      - split: validation - "data/pharmacy_val.csv"
      - split: test - "data/pharmacy_test.csv"
  - config_name: educational_psychology
    datafiles:
      - split: train - "data/educational_psychology_dev.csv"
      - split: validation - "data/educational_psychology_val.csv"
      - split: test - "data/educational_psychology_test.csv"
  - config_name: statistics_and_machine_learning
    datafiles:
      - split: train - "data/statistics_and_machine_learning_dev.csv"
      - split: validation - "data/statistics_and_machine_learning_val.csv"
      - split: test - "data/statistics_and_machine_learning_test.csv"
  - config_name: management_accounting
    datafiles:
      - split: train - "data/management_accounting_dev.csv"
      - split: validation - "data/management_accounting_val.csv"
      - split: test - "data/management_accounting_test.csv"
  - config_name: introduction_to_law
    datafiles:
      - split: train - "data/introduction_to_law_dev.csv"
      - split: validation - "data/introduction_to_law_val.csv"
      - split: test - "data/introduction_to_law_test.csv"
  - config_name: computer_science
    datafiles:
      - split: train - "data/computer_science_dev.csv"
      - split: validation - "data/computer_science_val.csv"
      - split: test - "data/computer_science_test.csv"
  - config_name: veterinary_pathology
    datafiles:
      - split: train - "data/veterinary_pathology_dev.csv"
      - split: validation - "data/veterinary_pathology_val.csv"
      - split: test - "data/veterinary_pathology_test.csv"
  - config_name: accounting
    datafiles:
      - split: train - "data/accounting_dev.csv"
      - split: validation - "data/accounting_val.csv"
      - split: test - "data/accounting_test.csv"
  - config_name: fire_science
    datafiles:
      - split: train - "data/fire_science_dev.csv"
      - split: validation - "data/fire_science_val.csv"
      - split: test - "data/fire_science_test.csv"
  - config_name: optometry
    datafiles:
      - split: train - "data/optometry_dev.csv"
      - split: validation - "data/optometry_val.csv"
      - split: test - "data/optometry_test.csv"
  - config_name: insurance_studies
    datafiles:
      - split: train - "data/insurance_studies_dev.csv"
      - split: validation - "data/insurance_studies_val.csv"
      - split: test - "data/insurance_studies_test.csv"
  - config_name: pharmacology
    datafiles:
      - split: train - "data/pharmacology_dev.csv"
      - split: validation - "data/pharmacology_val.csv"
      - split: test - "data/pharmacology_test.csv"
  - config_name: taxation
    datafiles:
      - split: train - "data/taxation_dev.csv"
      - split: validation - "data/taxation_val.csv"
      - split: test - "data/taxation_test.csv"
  - config_name: trust_practice
    datafiles:
      - split: train - "data/trust_practice_dev.csv"
      - split: validation - "data/trust_practice_val.csv"
      - split: test - "data/trust_practice_test.csv"
  - config_name: geography_of_taiwan
    datafiles:
      - split: train - "data/geography_of_taiwan_dev.csv"
      - split: validation - "data/geography_of_taiwan_val.csv"
      - split: test - "data/geography_of_taiwan_test.csv"
  - config_name: physical_education
    datafiles:
      - split: train - "data/physical_education_dev.csv"
      - split: validation - "data/physical_education_val.csv"
      - split: test - "data/physical_education_test.csv"
  - config_name: auditing
    datafiles:
      - split: train - "data/auditing_dev.csv"
      - split: validation - "data/auditing_val.csv"
      - split: test - "data/auditing_test.csv"
  - config_name: administrative_law
    datafiles:
      - split: train - "data/administrative_law_dev.csv"
      - split: validation - "data/administrative_law_val.csv"
      - split: test - "data/administrative_law_test.csv"
  - config_name: education_(profession_level)
    datafiles:
      - split: train - "data/education_(profession_level)_dev.csv"
      - split: validation - "data/education_(profession_level)_val.csv"
      - split: test - "data/education_(profession_level)_test.csv"
  - config_name: economics
    datafiles:
      - split: train - "data/economics_dev.csv"
      - split: validation - "data/economics_val.csv"
      - split: test - "data/economics_test.csv"
  - config_name: veterinary_pharmacology
    datafiles:
      - split: train - "data/veterinary_pharmacology_dev.csv"
      - split: validation - "data/veterinary_pharmacology_val.csv"
      - split: test - "data/veterinary_pharmacology_test.csv"
  - config_name: nautical_science
    datafiles:
      - split: train - "data/nautical_science_dev.csv"
      - split: validation - "data/nautical_science_val.csv"
      - split: test - "data/nautical_science_test.csv"
  - config_name: occupational_therapy_for_psychological_disorders
    datafiles:
      - split: >-
          train -
          "data/occupational_therapy_for_psychological_disorders_dev.csv"
      - split: >-
          validation -
          "data/occupational_therapy_for_psychological_disorders_val.csv"
      - split: >-
          test -
          "data/occupational_therapy_for_psychological_disorders_test.csv"
  - config_name: basic_medical_science
    datafiles:
      - split: train - "data/basic_medical_science_dev.csv"
      - split: validation - "data/basic_medical_science_val.csv"
      - split: test - "data/basic_medical_science_test.csv"
  - config_name: macroeconomics
    datafiles:
      - split: train - "data/macroeconomics_dev.csv"
      - split: validation - "data/macroeconomics_val.csv"
      - split: test - "data/macroeconomics_test.csv"
  - config_name: trade
    datafiles:
      - split: train - "data/trade_dev.csv"
      - split: validation - "data/trade_val.csv"
      - split: test - "data/trade_test.csv"
  - config_name: chinese_language_and_literature
    datafiles:
      - split: train - "data/chinese_language_and_literature_dev.csv"
      - split: validation - "data/chinese_language_and_literature_val.csv"
      - split: test - "data/chinese_language_and_literature_test.csv"
  - config_name: tve_design
    datafiles:
      - split: train - "data/tve_design_dev.csv"
      - split: validation - "data/tve_design_val.csv"
      - split: test - "data/tve_design_test.csv"
  - config_name: junior_science_exam
    datafiles:
      - split: train - "data/junior_science_exam_dev.csv"
      - split: validation - "data/junior_science_exam_val.csv"
      - split: test - "data/junior_science_exam_test.csv"
  - config_name: junior_math_exam
    datafiles:
      - split: train - "data/junior_math_exam_dev.csv"
      - split: validation - "data/junior_math_exam_val.csv"
      - split: test - "data/junior_math_exam_test.csv"
  - config_name: junior_chinese_exam
    datafiles:
      - split: train - "data/junior_chinese_exam_dev.csv"
      - split: validation - "data/junior_chinese_exam_val.csv"
      - split: test - "data/junior_chinese_exam_test.csv"
  - config_name: junior_social_studies
    datafiles:
      - split: train - "data/junior_social_studies_dev.csv"
      - split: validation - "data/junior_social_studies_val.csv"
      - split: test - "data/junior_social_studies_test.csv"
  - config_name: tve_mathematics
    datafiles:
      - split: train - "data/tve_mathematics_dev.csv"
      - split: validation - "data/tve_mathematics_val.csv"
      - split: test - "data/tve_mathematics_test.csv"
  - config_name: tve_chinese_language
    datafiles:
      - split: train - "data/tve_chinese_language_dev.csv"
      - split: validation - "data/tve_chinese_language_val.csv"
      - split: test - "data/tve_chinese_language_test.csv"
  - config_name: tve_natural_sciences
    datafiles:
      - split: train - "data/tve_natural_sciences_dev.csv"
      - split: validation - "data/tve_natural_sciences_val.csv"
      - split: test - "data/tve_natural_sciences_test.csv"
  - config_name: junior_chemistry
    datafiles:
      - split: train - "data/junior_chemistry_dev.csv"
      - split: validation - "data/junior_chemistry_val.csv"
      - split: test - "data/junior_chemistry_test.csv"
  - config_name: music
    datafiles:
      - split: train - "data/music_dev.csv"
      - split: validation - "data/music_val.csv"
      - split: test - "data/music_test.csv"
  - config_name: education
    datafiles:
      - split: train - "data/education_dev.csv"
      - split: validation - "data/education_val.csv"
      - split: test - "data/education_test.csv"
  - config_name: three_principles_of_people
    datafiles:
      - split: train - "data/three_principles_of_people_dev.csv"
      - split: validation - "data/three_principles_of_people_val.csv"
      - split: test - "data/three_principles_of_people_test.csv"
  - config_name: taiwanese_hokkien
    datafiles:
      - split: train - "data/taiwanese_hokkien_dev.csv"
      - split: validation - "data/taiwanese_hokkien_val.csv"
      - split: test - "data/taiwanese_hokkien_test.csv"

TMMLU+ : Large scale traditional chinese massive multitask language understanding

A close-up image of a neat paper note with a white background. The text 'TMMLU+' is written horizontally across the center of the note in bold, black. Join us to work in multimodal LLM : https://ikala.ai/recruit/

We present TMMLU+, a traditional Chinese massive multitask language understanding dataset. TMMLU+ is a multiple-choice question-answering dataset featuring 66 subjects, ranging from elementary to professional level.

The TMMLU+ dataset is six times larger and contains more balanced subjects compared to its predecessor, TMMLU. We have included benchmark results in TMMLU+ from closed-source models and 20 open-source Chinese large language models, with parameters ranging from 1.8B to 72B. The benchmark results show that Traditional Chinese variants still lag behind those trained on major Simplified Chinese models.

from datasets import load_dataset
task_list = [
             'engineering_math', 'dentistry', 'traditional_chinese_medicine_clinical_medicine', 'clinical_psychology', 'technical', 'culinary_skills', 'mechanical', 'logic_reasoning', 'real_estate',
             'general_principles_of_law', 'finance_banking', 'anti_money_laundering', 'ttqav2', 'marketing_management', 'business_management', 'organic_chemistry', 'advance_chemistry',
             'physics', 'secondary_physics', 'human_behavior', 'national_protection', 'jce_humanities', 'politic_science', 'agriculture', 'official_document_management',
             'financial_analysis', 'pharmacy', 'educational_psychology', 'statistics_and_machine_learning', 'management_accounting', 'introduction_to_law', 'computer_science', 'veterinary_pathology',
             'accounting', 'fire_science', 'optometry', 'insurance_studies', 'pharmacology', 'taxation', 'trust_practice', 'geography_of_taiwan', 'physical_education', 'auditing', 'administrative_law',
             'education_(profession_level)', 'economics', 'veterinary_pharmacology', 'nautical_science', 'occupational_therapy_for_psychological_disorders',
             'basic_medical_science', 'macroeconomics', 'trade', 'chinese_language_and_literature', 'tve_design', 'junior_science_exam', 'junior_math_exam', 'junior_chinese_exam',
             'junior_social_studies', 'tve_mathematics', 'tve_chinese_language', 'tve_natural_sciences', 'junior_chemistry', 'music', 'education', 'three_principles_of_people',
             'taiwanese_hokkien'
            ]
for task in task_list:
  val = load_dataset('ikala/tmmluplus', task)['validation']
  dev = load_dataset('ikala/tmmluplus', task)['train']
  test = load_dataset('ikala/tmmluplus', task)['test']

Statistic on all four categories : STEM, Social Science, Humanities, Other

Category	Test	Dev	Validation
STEM	3458	70	385
Social Sciences	5958	90	665
Humanities	1763	35	197
Other (Business, Health, Misc.)	8939	135	995
Total	20118	330	2242

Benchmark on direct prompting

model	STEM	Social Science	Humanities	Other	Average
Qwen/Qwen-72B	61.12	71.65	63.00	61.31	64.27
Qwen/Qwen-14B	46.94	56.69	49.43	48.81	50.47
Gemini-pro	45.38	57.29	48.80	48.21	49.92
01-ai/Yi-34B-Chat	40.24	56.77	53.99	47.58	49.64
Qwen/Qwen-14B-Chat	43.86	53.29	44.78	45.13	46.77
01-ai/Yi-6B-Chat	39.62	50.24	44.44	44.26	44.64
Claude-1.3	42.65	49.33	42.16	44.14	44.57
gpt-3.5-turbo-0613	41.56	46.72	36.73	42.03	41.76
CausalLM/14B	39.83	44.50	39.61	41.97	41.48
Skywork/Skywork-13B-base	36.93	47.27	41.04	40.10	41.33
Qwen/Qwen-7B	37.53	45.48	38.09	38.96	40.01
Qwen/Qwen-7B-Chat	33.32	44.64	40.27	39.89	39.53
vivo-ai/BlueLM-7B-Base	33.94	41.52	37.38	38.74	37.90
baichuan-inc/Baichuan2-13B-Chat	29.64	43.73	37.36	39.88	37.65
Qwen/Qwen-1_8B	32.65	38.95	38.34	35.27	36.30
Claude-2	39.65	39.09	28.59	37.47	36.20
THUDM/chatglm3-6b	31.05	39.31	35.64	35.60	35.40
deepseek-ai/deepseek-llm-7b-chat	29.82	42.29	34.24	34.31	35.17
CausalLM/7B	31.03	38.17	35.87	35.39	35.11
Azure99/blossom-v3_1-mistral-7b	32.80	36.91	32.36	34.53	34.15
Qwen/Qwen-1_8B-Chat	26.60	36.36	31.81	31.96	31.68
TigerResearch/tigerbot-13b-chat-v3	24.73	29.63	25.72	27.22	26.82
hongyin/mistral-7b-80k	24.26	23.76	22.56	24.57	23.79
yentinglin/Taiwan-LLM-13B-v2.0-chat	18.53	27.65	17.77	21.49	21.36
LinkSoul/Chinese-Llama-2-7b	16.55	18.39	12.97	16.13	16.01
yentinglin/Taiwan-LLM-7B-v2.1-chat	14.99	16.23	15.00	16.22	15.61
FlagAlpha/Atom-7B	5.60	13.57	7.71	11.84	9.68