| from __future__ import annotations |
|
|
| from typing import Generator |
|
|
| import datasets |
|
|
| from .classification import ( |
| AmazonCounterfactualClassificationConfig, |
| AmazonCounterfactualClassificationHelper, |
| AmazonReviewClassificationConfig, |
| AmazonReviewClassificationHelper, |
| MassiveIntentClassificationConfig, |
| MassiveIntentClassificationHelper, |
| MassiveScenarioClassificationConfig, |
| MassiveScenarioClassificationHelper, |
| ) |
| from .clustering import ( |
| LivedoorNewsConfig, |
| LivedoorNewsHelper, |
| MewsC16JaConfig, |
| MewsC16JaHelper, |
| ) |
| from .pair_classification import PAWSXJaConfig, PAWSXJaHelper |
| from .reranking import ( |
| EsciConfig, |
| EsciHelper, |
| JaCWIRRerankingConfig, |
| JaCWIRRerankingHelper, |
| JQaRAConfig, |
| JQaRAHelper, |
| MIRACLRerankingConfig, |
| MIRACLRerankingHelper, |
| MLDRRerankingConfig, |
| MLDRRerankingHelper, |
| ) |
| from .retrieval import ( |
| JaCWIRRetrievalConfig, |
| JaCWIRRetrievalHelper, |
| JaGovFaqs22kConfig, |
| JaGovFaqs22kHelper, |
| JAQKETConfig, |
| JAQKETHelper, |
| MIRACLRetrievalConfig, |
| MIRACLRetrievalHelper, |
| MLDRRetrievalConfig, |
| MLDRRetrievalHelper, |
| MrTyDiConfig, |
| MrTyDiHelper, |
| NLPJournalConfig, |
| NLPJournalHelper, |
| ) |
| from .sts import JSICKConfig, JSICKHelper, JSTSConfig, JSTSHelper |
|
|
|
|
| class JMTEB(datasets.GeneratorBasedBuilder): |
| BUILDER_CONFIGS = [ |
| LivedoorNewsConfig(), |
| MewsC16JaConfig(), |
| AmazonReviewClassificationConfig(), |
| AmazonCounterfactualClassificationConfig(), |
| MassiveIntentClassificationConfig(), |
| MassiveScenarioClassificationConfig(), |
| JSTSConfig(), |
| JSICKConfig(), |
| PAWSXJaConfig(), |
| JAQKETConfig(name="jaqket-query"), |
| JAQKETConfig(name="jaqket-corpus"), |
| MrTyDiConfig(name="mrtydi-query"), |
| MrTyDiConfig(name="mrtydi-corpus"), |
| NLPJournalConfig(name="nlp_journal_title_abs-query"), |
| NLPJournalConfig(name="nlp_journal_title_abs-corpus"), |
| NLPJournalConfig(name="nlp_journal_title_intro-query"), |
| NLPJournalConfig(name="nlp_journal_title_intro-corpus"), |
| NLPJournalConfig(name="nlp_journal_abs_intro-query"), |
| NLPJournalConfig(name="nlp_journal_abs_intro-corpus"), |
| NLPJournalConfig(name="nlp_journal_abs_article-query"), |
| NLPJournalConfig(name="nlp_journal_abs_article-corpus"), |
| JaGovFaqs22kConfig(name="jagovfaqs_22k-query"), |
| JaGovFaqs22kConfig(name="jagovfaqs_22k-corpus"), |
| JaCWIRRetrievalConfig(name="jacwir-retrieval-query"), |
| JaCWIRRetrievalConfig(name="jacwir-retrieval-corpus"), |
| MIRACLRetrievalConfig(name="miracl-retrieval-query"), |
| MIRACLRetrievalConfig(name="miracl-retrieval-corpus"), |
| MLDRRetrievalConfig(name="mldr-retrieval-query"), |
| MLDRRetrievalConfig(name="mldr-retrieval-corpus"), |
| EsciConfig(name="esci-query"), |
| EsciConfig(name="esci-corpus"), |
| JQaRAConfig(name="jqara-query"), |
| JQaRAConfig(name="jqara-corpus"), |
| JaCWIRRerankingConfig(name="jacwir-reranking-query"), |
| JaCWIRRerankingConfig(name="jacwir-reranking-corpus"), |
| MIRACLRerankingConfig(name="miracl-reranking-query"), |
| MIRACLRerankingConfig(name="miracl-reranking-corpus"), |
| MLDRRerankingConfig(name="mldr-reranking-query"), |
| MLDRRerankingConfig(name="mldr-reranking-corpus"), |
| ] |
|
|
| def _info(self) -> datasets.DatasetInfo: |
| |
| if self.config.name == "livedoor_news": |
| return datasets.DatasetInfo( |
| description=LivedoorNewsHelper.DESCRIPTION, |
| citation=LivedoorNewsHelper.CITATION, |
| homepage=LivedoorNewsHelper.HOMEPAGE, |
| license=LivedoorNewsHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "url": datasets.Value("string"), |
| "timestamp": datasets.Value("string"), |
| "title": datasets.Value("string"), |
| "text": datasets.Value("string"), |
| "label": datasets.features.ClassLabel( |
| names=LivedoorNewsHelper.CATEGORIES |
| ), |
| } |
| ), |
| ) |
| if self.config.name == "mewsc16_ja": |
| return datasets.DatasetInfo( |
| description=MewsC16JaHelper.DESCRIPTION, |
| citation=MewsC16JaHelper.CITATION, |
| homepage=MewsC16JaHelper.HOMEPAGE, |
| license=MewsC16JaHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "idx": datasets.Value("int32"), |
| "text": datasets.Value("string"), |
| "label": datasets.features.ClassLabel( |
| names=MewsC16JaHelper.CATEGORIES |
| ), |
| } |
| ), |
| ) |
|
|
| |
| if self.config.name == "amazon_review_classification": |
| return datasets.DatasetInfo( |
| description=AmazonReviewClassificationHelper.DESCRIPTION, |
| citation=AmazonReviewClassificationHelper.CITATION, |
| homepage=AmazonReviewClassificationHelper.HOMEPAGE_URL, |
| license=AmazonReviewClassificationHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "id": datasets.Value("string"), |
| "text": datasets.Value("string"), |
| "label": datasets.Value("int32"), |
| "label_text": datasets.Value("string"), |
| } |
| ), |
| ) |
| if self.config.name == "amazon_counterfactual_classification": |
| return datasets.DatasetInfo( |
| description=AmazonCounterfactualClassificationHelper.DESCRIPTION, |
| citation=AmazonCounterfactualClassificationHelper.CITATION, |
| homepage=AmazonCounterfactualClassificationHelper.HOMEPAGE_URL, |
| license=AmazonCounterfactualClassificationHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "text": datasets.Value("string"), |
| "label": datasets.Value("int32"), |
| "label_text": datasets.Value("string"), |
| } |
| ), |
| ) |
| if self.config.name == "massive_intent_classification": |
| return datasets.DatasetInfo( |
| description=MassiveIntentClassificationHelper.DESCRIPTION, |
| citation=MassiveIntentClassificationHelper.CITATION, |
| homepage=MassiveIntentClassificationHelper.HOMEPAGE_URL, |
| license=MassiveIntentClassificationHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "id": datasets.Value("string"), |
| "label": datasets.features.ClassLabel( |
| names=MassiveIntentClassificationHelper.INTENTS |
| ), |
| "label_text": datasets.Value("string"), |
| "text": datasets.Value("string"), |
| }, |
| ), |
| ) |
| if self.config.name == "massive_scenario_classification": |
| return datasets.DatasetInfo( |
| description=MassiveScenarioClassificationHelper.DESCRIPTION, |
| citation=MassiveScenarioClassificationHelper.CITATION, |
| homepage=MassiveScenarioClassificationHelper.HOMEPAGE_URL, |
| license=MassiveScenarioClassificationHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "id": datasets.Value("string"), |
| "label": datasets.features.ClassLabel( |
| names=MassiveScenarioClassificationHelper.SCENARIOS |
| ), |
| "label_text": datasets.Value("string"), |
| "text": datasets.Value("string"), |
| }, |
| ), |
| ) |
|
|
| |
| if self.config.name == "jsts": |
| return datasets.DatasetInfo( |
| description=JSTSHelper.DESCRIPTION, |
| citation=JSTSHelper.CITATION, |
| homepage=JSTSHelper.HOMEPAGE_URL, |
| license=JSTSHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "sentence_pair_id": datasets.Value("string"), |
| "yjcaptions_id": datasets.Value("string"), |
| "sentence1": datasets.Value("string"), |
| "sentence2": datasets.Value("string"), |
| "label": datasets.Value("float"), |
| }, |
| ), |
| ) |
| if self.config.name == "jsick": |
| return datasets.DatasetInfo( |
| description=JSICKHelper.DESCRIPTION, |
| citation=JSICKHelper.CITATION, |
| homepage=JSTSHelper.HOMEPAGE_URL, |
| license=JSTSHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "idx": datasets.Value("int32"), |
| "sentence1": datasets.Value("string"), |
| "sentence2": datasets.Value("string"), |
| "label": datasets.Value("float"), |
| }, |
| ), |
| ) |
|
|
| |
| if self.config.name == "paws_x_ja": |
| return datasets.DatasetInfo( |
| description=PAWSXJaHelper.DESCRIPTION, |
| citation=PAWSXJaHelper.CITATION, |
| homepage=PAWSXJaHelper.HOMEPAGE_URL, |
| license=PAWSXJaHelper.LICENSE, |
| features=datasets.Features( |
| { |
| "id": datasets.Value("int32"), |
| "sentence1": datasets.Value("string"), |
| "sentence2": datasets.Value("string"), |
| "label": datasets.Value("int32"), |
| }, |
| ), |
| ) |
|
|
| |
| if self.config.name.startswith("jaqket"): |
| return datasets.DatasetInfo( |
| description=JAQKETHelper.DESCRIPTION, |
| citation=JAQKETHelper.CITATION, |
| homepage=JAQKETHelper.HOMEPAGE_URL, |
| license=JAQKETHelper.LICENSE, |
| ) |
| if self.config.name.startswith("mrtydi"): |
| return datasets.DatasetInfo( |
| description=MrTyDiHelper.DESCRIPTION, |
| citation=MrTyDiHelper.CITATION, |
| homepage=MrTyDiHelper.HOMEPAGE_URL, |
| license=MrTyDiHelper.LICENSE, |
| ) |
| if self.config.name.startswith("nlp_journal"): |
| return datasets.DatasetInfo( |
| description=NLPJournalHelper.DESCRIPTION, |
| citation=NLPJournalHelper.CITATION, |
| homepage=NLPJournalHelper.HOMEPAGE_URL, |
| license=NLPJournalHelper.LICENSE, |
| ) |
| if self.config.name.startswith("jagovfaqs_22k"): |
| return datasets.DatasetInfo( |
| description=JaGovFaqs22kHelper.DESCRIPTION, |
| citation=JaGovFaqs22kHelper.CITATION, |
| homepage=JaGovFaqs22kHelper.HOMEPAGE_URL, |
| license=JaGovFaqs22kHelper.LICENSE, |
| ) |
| if self.config.name.startswith("jacwir-retrieval"): |
| return datasets.DatasetInfo( |
| description=JaCWIRRetrievalHelper.DESCRIPTION, |
| citation=JaCWIRRetrievalHelper.CITATION, |
| homepage=JaCWIRRetrievalHelper.HOMEPAGE_URL, |
| license=JaCWIRRetrievalHelper.LICENSE, |
| ) |
| if self.config.name.startswith("miracl-retrieval"): |
| return datasets.DatasetInfo( |
| description=MIRACLRetrievalHelper.DESCRIPTION, |
| citation=MIRACLRetrievalHelper.CITATION, |
| homepage=MIRACLRetrievalHelper.HOMEPAGE_URL, |
| license=MIRACLRetrievalHelper.LICENSE, |
| ) |
| if self.config.name.startswith("mldr-retrieval"): |
| return datasets.DatasetInfo( |
| description=MLDRRetrievalHelper.DESCRIPTION, |
| citation=MLDRRetrievalHelper.CITATION, |
| homepage=MLDRRetrievalHelper.HOMEPAGE_URL, |
| license=MLDRRetrievalHelper.LICENSE, |
| ) |
|
|
| |
| if self.config.name.startswith("esci"): |
| return datasets.DatasetInfo( |
| description=EsciHelper.DESCRIPTION, |
| citation=EsciHelper.CITATION, |
| homepage=EsciHelper.HOMEPAGE_URL, |
| license=EsciHelper.LICENSE, |
| ) |
| if self.config.name.startswith("jqara"): |
| return datasets.DatasetInfo( |
| description=JQaRAHelper.DESCRIPTION, |
| citation=JQaRAHelper.CITATION, |
| homepage=JQaRAHelper.HOMEPAGE_URL, |
| license=JQaRAHelper.LICENSE, |
| ) |
| if self.config.name.startswith("jacwir-reranking"): |
| return datasets.DatasetInfo( |
| description=JaCWIRRerankingHelper.DESCRIPTION, |
| citation=JaCWIRRerankingHelper.CITATION, |
| homepage=JaCWIRRerankingHelper.HOMEPAGE_URL, |
| license=JaCWIRRerankingHelper.LICENSE, |
| ) |
| if self.config.name.startswith("miracl-reranking"): |
| return datasets.DatasetInfo( |
| description=MIRACLRerankingHelper.DESCRIPTION, |
| citation=MIRACLRerankingHelper.CITATION, |
| homepage=MIRACLRerankingHelper.HOMEPAGE_URL, |
| license=MIRACLRerankingHelper.LICENSE, |
| ) |
| if self.config.name.startswith("mldr-reranking"): |
| return datasets.DatasetInfo( |
| description=MLDRRerankingHelper.DESCRIPTION, |
| citation=MLDRRerankingHelper.CITATION, |
| homepage=MLDRRerankingHelper.HOMEPAGE_URL, |
| license=MLDRRerankingHelper.LICENSE, |
| ) |
|
|
| def _split_generators( |
| self, dl_manager: datasets.DownloadManager |
| ) -> list[datasets.SplitGenerator]: |
| |
| if self.config.name == "livedoor_news": |
| config: LivedoorNewsConfig = self.config |
| livedoor_news_helper = LivedoorNewsHelper() |
| return livedoor_news_helper.split_generator(config, dl_manager) |
| if self.config.name == "mewsc16_ja": |
| return MewsC16JaHelper().split_generator(self.config, dl_manager) |
|
|
| |
| if self.config.name == "amazon_review_classification": |
| return AmazonReviewClassificationHelper.split_generator() |
| if self.config.name == "amazon_counterfactual_classification": |
| return AmazonCounterfactualClassificationHelper.split_generator() |
| if self.config.name == "massive_intent_classification": |
| return MassiveIntentClassificationHelper.split_generator() |
| if self.config.name == "massive_scenario_classification": |
| return MassiveScenarioClassificationHelper.split_generator() |
|
|
| |
| if self.config.name == "jsts": |
| config: JSTSConfig = self.config |
| jsts_helper = JSTSHelper() |
| return jsts_helper.split_generator(dl_manager) |
| if self.config.name == "jsick": |
| config: JSICKConfig = self.config |
| jsick_helper = JSICKHelper() |
| return jsick_helper.split_generator(config, dl_manager) |
|
|
| |
| if self.config.name == "paws_x_ja": |
| return PAWSXJaHelper.split_generator() |
|
|
| |
| if self.config.name == "jaqket-query": |
| return JAQKETHelper().query_split_generator(dl_manager) |
| if self.config.name == "jaqket-corpus": |
| return JAQKETHelper().corpus_generator(dl_manager) |
| if self.config.name == "mrtydi-query": |
| return MrTyDiHelper.query_split_generator() |
| if self.config.name == "mrtydi-corpus": |
| return MrTyDiHelper.corpus_generator() |
|
|
| if self.config.name.startswith("nlp_journal"): |
| nlp_journal_helper = NLPJournalHelper(self.config, dl_manager) |
| if self.config.name == "nlp_journal_title_abs-query": |
| return nlp_journal_helper.query_generator("title", "abstract") |
| if self.config.name == "nlp_journal_title_abs-corpus": |
| return nlp_journal_helper.corpus_generator("title", "abstract") |
| if self.config.name == "nlp_journal_title_intro-query": |
| return nlp_journal_helper.query_generator("title", "introduction") |
| if self.config.name == "nlp_journal_title_intro-corpus": |
| return nlp_journal_helper.corpus_generator("title", "introduction") |
| if self.config.name == "nlp_journal_abs_intro-query": |
| return nlp_journal_helper.query_generator("abstract", "introduction") |
| if self.config.name == "nlp_journal_abs_intro-corpus": |
| return nlp_journal_helper.corpus_generator("abstract", "introduction") |
| if self.config.name == "nlp_journal_abs_article-query": |
| return nlp_journal_helper.query_generator("abstract", "article") |
| if self.config.name == "nlp_journal_abs_article-corpus": |
| return nlp_journal_helper.corpus_generator("abstract", "article") |
|
|
| if self.config.name.startswith("jagovfaqs_22k"): |
| jagovfaq_22k_helper = JaGovFaqs22kHelper(self.config) |
| if self.config.name == "jagovfaqs_22k-query": |
| return jagovfaq_22k_helper.query_split_generator() |
| if self.config.name == "jagovfaqs_22k-corpus": |
| return jagovfaq_22k_helper.corpus_generator() |
|
|
| if self.config.name.startswith("jacwir-retrieval"): |
| jacwir_retrieval_helper = JaCWIRRetrievalHelper(self.config) |
| if self.config.name == "jacwir-retrieval-query": |
| return jacwir_retrieval_helper.query_split_generator() |
| if self.config.name == "jacwir-retrieval-corpus": |
| return jacwir_retrieval_helper.corpus_generator() |
|
|
| if self.config.name.startswith("miracl-retrieval"): |
| miracl_retrieval_helper = MIRACLRetrievalHelper(self.config) |
| if self.config.name == "miracl-retrieval-query": |
| return miracl_retrieval_helper.query_split_generator() |
| if self.config.name == "miracl-retrieval-corpus": |
| return miracl_retrieval_helper.corpus_generator() |
|
|
| if self.config.name.startswith("mldr-retrieval"): |
| mldr_retrieval_helper = MLDRRetrievalHelper(self.config) |
| if self.config.name == "mldr-retrieval-query": |
| return mldr_retrieval_helper.query_split_generator() |
| if self.config.name == "mldr-retrieval-corpus": |
| return mldr_retrieval_helper.corpus_generator() |
|
|
| |
| if self.config.name.startswith("esci"): |
| esci_helper = EsciHelper(self.config) |
| if self.config.name == "esci-query": |
| return esci_helper.query_split_generator() |
| if self.config.name == "esci-corpus": |
| return esci_helper.corpus_generator() |
|
|
| if self.config.name.startswith("jqara"): |
| jqara_helper = JQaRAHelper(self.config) |
| if self.config.name == "jqara-query": |
| return jqara_helper.query_split_generator() |
| if self.config.name == "jqara-corpus": |
| return jqara_helper.corpus_generator() |
|
|
| if self.config.name.startswith("jacwir-reranking"): |
| jacwir_reranking_helper = JaCWIRRerankingHelper(self.config) |
| if self.config.name == "jacwir-reranking-query": |
| return jacwir_reranking_helper.query_split_generator() |
| if self.config.name == "jacwir-reranking-corpus": |
| return jacwir_reranking_helper.corpus_generator() |
|
|
| if self.config.name.startswith("miracl-reranking"): |
| miracl_reranking_helper = MIRACLRerankingHelper(self.config) |
| if self.config.name == "miracl-reranking-query": |
| return miracl_reranking_helper.query_split_generator() |
| if self.config.name == "miracl-reranking-corpus": |
| return miracl_reranking_helper.corpus_generator() |
|
|
| if self.config.name.startswith("mldr-reranking"): |
| mldr_reranking_helper = MLDRRerankingHelper(self.config) |
| if self.config.name == "mldr-reranking-query": |
| return mldr_reranking_helper.query_split_generator() |
| if self.config.name == "mldr-reranking-corpus": |
| return mldr_reranking_helper.corpus_generator() |
|
|
| def _generate_examples(self, data: datasets.SplitGenerator) -> Generator: |
| for i, d in enumerate(data): |
| yield i, d |
|
|
