__init__.py

from pprint import pprint

import sacrebleu

from . import superglue
from . import glue
from . import arc
from . import coqa
from . import race
from . import webqs
from . import anli
from . import wsc273
from . import winogrande
from . import quac
from . import hellaswag
from . import openbookqa
from . import squad
from . import naturalqs
from . import sat
from . import arithmetic
from . import lambada
from . import race
from . import piqa
from . import prost
from . import triviaqa
from . import pubmedqa
from . import sciq
from . import webqs
from . import qa4mre
from . import translation
from . import headqa
from . import mathqa
from . import hendrycks_ethics
from . import drop
from . import unscramble
from . import logiqa
from . import hendrycks_test
from . import hendrycks_math
from . import cbt
from . import lambada_cloze
from . import pile
from . import wikitext
from . import xquad

########################################
# Translation tasks
########################################

# 6 total
gpt3_translation_benchmarks = {
    "wmt14": ['en-fr', 'fr-en'],  # French
    "wmt16": ['en-ro', 'ro-en', 'de-en', 'en-de'],  # German, Romanian
}


# 28 total
selected_translation_benchmarks = {
    **gpt3_translation_benchmarks,
    "wmt20": sacrebleu.get_langpairs_for_testset("wmt20"),
    "iwslt17": ['en-ar', 'ar-en']  # Arabic
}

# 319 total
all_translation_benchmarks = {
    ts: sacrebleu.get_langpairs_for_testset(ts)
    for ts in sacrebleu.get_available_testsets()
}


########################################
# All tasks
########################################


TASK_REGISTRY = {
    # GLUE
    "cola": glue.CoLA,
    "mnli": glue.MNLI,
    "mnli_mismatched": glue.MNLIMismatched,
    "mrpc": glue.MRPC,
    "rte": glue.RTE,
    "qnli": glue.QNLI,
    "qqp": glue.QQP,
    #"stsb": glue.STSB, # not implemented yet
    "sst": glue.SST,
    "wnli": glue.WNLI,
    # SuperGLUE
    "boolq": superglue.BoolQ,
    "cb": superglue.CommitmentBank,
    "copa": superglue.Copa,
    "multirc": superglue.MultiRC,
    "record": superglue.ReCoRD,
    "wic": superglue.WordsInContext,
    "wsc": superglue.SGWinogradSchemaChallenge,
    
    # Order by benchmark/genre?
    "coqa": coqa.CoQA,
    "drop": drop.DROP,
    "lambada": lambada.LAMBADA,
    "lambada_cloze": lambada_cloze.LAMBADA_cloze,
    "wikitext": wikitext.WikiText,
    # "cbt-cn": cbt.CBTCN, # disabled pending context length fix
    # "cbt-ne": cbt.CBTNE, # disabled pending context length fix

    "piqa": piqa.PiQA,
    "prost": prost.PROST,

    # Science related
    "pubmedqa" : pubmedqa.Pubmed_QA,
    "sciq" : sciq.SciQ,

    "qa4mre_2011" : qa4mre.QA4MRE_2011,
    "qa4mre_2012" : qa4mre.QA4MRE_2012,
    "qa4mre_2013" : qa4mre.QA4MRE_2013,

    "triviaqa": triviaqa.TriviaQA,
    "arc_easy": arc.ARCEasy,
    "arc_challenge": arc.ARCChallenge,
    # "quac": quac.QuAC, # not implemented yet
    "logiqa": logiqa.LogiQA,
    "hellaswag": hellaswag.HellaSwag,
    "openbookqa": openbookqa.OpenBookQA,
    # "sat": sat.SATAnalogies, # not implemented yet
    "squad2": squad.SQuAD2,

    # xquad tasks
    "xquad_ar": xquad.XQuADAr,
    "xquad_de": xquad.XQuADDe,
    "xquad_zh": xquad.XQuADZh,
    "xquad_vi": xquad.XQuADVi,
    "xquad_en": xquad.XQuADEn,
    "xquad_es": xquad.XQuADEs,
    "xquad_hi": xquad.XQuADHi,
    "xquad_el": xquad.XQuADEl,
    "xquad_th": xquad.XQuADTh,
    "xquad_tr": xquad.XQuADTr,
    "xquad_ru": xquad.XQuADRu,
    "xquad_ro": xquad.XQuADRo,

    "race": race.RACE,
    # "naturalqs": naturalqs.NaturalQs, # not implemented yet
    "headqa": headqa.HeadQA,
    "mathqa": mathqa.MathQA,
    "webqs": webqs.WebQs,
    "wsc273": wsc273.WinogradSchemaChallenge273,
    "winogrande": winogrande.Winogrande,
    "anli_r1": anli.ANLIRound1,
    "anli_r2": anli.ANLIRound2,
    "anli_r3": anli.ANLIRound3,

    "ethics_cm": hendrycks_ethics.EthicsCM,
    "ethics_deontology": hendrycks_ethics.EthicsDeontology,
    "ethics_justice": hendrycks_ethics.EthicsJustice,
    "ethics_utilitarianism_original": hendrycks_ethics.EthicsUtilitarianismOriginal,
    "ethics_utilitarianism": hendrycks_ethics.EthicsUtilitarianism,
    "ethics_virtue": hendrycks_ethics.EthicsVirtue,

    # math
    "math_algebra": hendrycks_math.MathAlgebra,
    "math_counting_and_prob": hendrycks_math.MathCountingAndProbability,
    "math_geometry": hendrycks_math.MathGeometry,
    "math_intermediate_algebra": hendrycks_math.MathIntermediateAlgebra,
    "math_num_theory": hendrycks_math.MathNumberTheory,
    "math_prealgebra": hendrycks_math.MathPrealgebra,
    "math_precalc": hendrycks_math.MathPrecalculus,

    # arithmetic
    "arithmetic_2da": arithmetic.Arithmetic2DPlus,
    "arithmetic_2ds": arithmetic.Arithmetic2DMinus,
    "arithmetic_3da": arithmetic.Arithmetic3DPlus,
    "arithmetic_3ds": arithmetic.Arithmetic3DMinus,
    "arithmetic_4da": arithmetic.Arithmetic4DPlus,
    "arithmetic_4ds": arithmetic.Arithmetic4DMinus,
    "arithmetic_5da": arithmetic.Arithmetic5DPlus,
    "arithmetic_5ds": arithmetic.Arithmetic5DMinus,
    "arithmetic_2dm": arithmetic.Arithmetic2DMultiplication,
    "arithmetic_1dc": arithmetic.Arithmetic1DComposite,
    # TODO Perhaps make these groups of tasks
    #   e.g. anli, arithmetic, openai_translations, harness_translations

    # hendrycksTest (57 tasks)
    **hendrycks_test.create_all_tasks(),

    # e.g. wmt14-fr-en
    **translation.create_tasks_from_benchmarks(gpt3_translation_benchmarks),
    # chef's selection, mostly wmt20
    **translation.create_tasks_from_benchmarks(selected_translation_benchmarks),

    # Word Scrambling and Manipulation Tasks
    "anagrams1": unscramble.Anagrams1,
    "anagrams2": unscramble.Anagrams2,
    "cycle_letters": unscramble.CycleLetters,
    "random_insertion": unscramble.RandomInsertion,
    "reversed_words": unscramble.ReversedWords,

    # Pile
    "pile_arxiv": pile.PileArxiv,
    "pile_books3": pile.PileBooks3,
    "pile_bookcorpus2": pile.PileBookCorpus2,
    "pile_dm-mathematics": pile.PileDmMathematics,
    "pile_enron": pile.PileEnron,
    "pile_europarl": pile.PileEuroparl,
    "pile_freelaw": pile.PileFreeLaw,
    "pile_github": pile.PileGithub,
    "pile_gutenberg": pile.PileGutenberg,
    "pile_hackernews": pile.PileHackernews,
    "pile_nih-exporter": pile.PileNIHExporter,
    "pile_opensubtitles": pile.PileOpenSubtitles,
    "pile_openwebtext2": pile.PileOpenWebText2,
    "pile_philpapers": pile.PilePhilPapers,
    "pile_pile-cc": pile.PilePileCc,
    "pile_pubmed-abstracts": pile.PilePubmedAbstracts,
    "pile_pubmed-central": pile.PilePubmedCentral,
    "pile_stackexchange": pile.PileStackExchange,
    "pile_uspto": pile.PileUspto,
    "pile_ubuntu-irc": pile.PileUbuntuIrc,
    "pile_wikipedia": pile.PileWikipedia,
    "pile_youtubesubtitles": pile.PileYoutubeSubtitles,
}


ALL_TASKS = sorted(list(TASK_REGISTRY))


def get_task(task_name):
    try:
        return TASK_REGISTRY[task_name]
    except KeyError as e:
        print("Available tasks:")
        pprint(TASK_REGISTRY)
        raise KeyError(f"Missing task {task_name}")


def get_task_dict(task_name_list):
    return {
        task_name: get_task(task_name)()
        for task_name in task_name_list
    }