tasks_config.yaml

version: 0.0.4
tasks:
  oab_exams:
    benchmark: oab_exams
    col_name: OAB Exams
    task_list:
    - oab_exams_generate
    metric: exact_match
    few_shot: 5
    limit: null
    baseline: 25.0
    human_baseline: 50.0
    description: OAB Exams is a dataset of 2,000 questions from the Brazilian Bar
      Association's exams.
    link: https://huggingface.co/datasets/eduagarcia/oab_exams
  brazilian_court_decisions_judgment:
    benchmark: brazilian_court_decisions_judgment
    col_name: BR Court Decisions
    task_list:
    - brazilian_court_decisions_judgment_generate
    metric: f1_macro
    few_shot: 5
    limit: null
    baseline: 33.33
    human_baseline: 100.0
    description: A classification dataset of court decisions from the Tribunal de
      Justiça de Alagoas (TJAL, the State Supreme Court of Alagoas (Brazil).
    link: https://huggingface.co/datasets/joelniklaus/brazilian_court_decisions
  datalawyer_frases:
    benchmark: datalawyer_frases
    col_name: DL Frases
    task_list:
    - datalawyer_frases_generate
    metric: f1_macro
    few_shot: 15
    limit: 2000
    baseline: 10.0
    human_baseline: 100.0
    description: A classification dataset
    link: https://huggingface.co/datasets/eduagarcia/portuguese_benchmark
  rrip:
    benchmark: rrip
    col_name: RRIP
    task_list:
    - rrip_generate
    metric: f1_macro
    few_shot: 15
    limit: null
    baseline: 12.5
    human_baseline: 100.0
    description: A classification dataset
    link: https://huggingface.co/datasets/eduagarcia/portuguese_benchmark