joya

  • Descripción :

GEM es un entorno de referencia para la generación de lenguaje natural con un enfoque en su evaluación, tanto a través de anotaciones humanas como de métricas automatizadas.

GEM tiene como objetivo: (1) medir el progreso de NLG en 13 conjuntos de datos que abarcan muchas tareas e idiomas de NLG. (2) proporcionar un análisis en profundidad de los datos y modelos presentados a través de declaraciones de datos y conjuntos de desafíos. (3) desarrollar estándares para la evaluación del texto generado usando métricas tanto automatizadas como humanas.

Se puede encontrar más información en https://gem-benchmark.com .

gem/common_gen (configuración predeterminada)

  • Descripción de la configuración : CommonGen es una tarea de generación de texto restringida, asociada con un conjunto de datos de referencia, para probar explícitamente las máquinas en cuanto a la capacidad de razonamiento generativo de sentido común. Dado un conjunto de conceptos comunes; la tarea es generar una oración coherente que describa un escenario cotidiano usando estos conceptos.

  • Tamaño de descarga : 1.84 MiB

  • Tamaño del conjunto de datos : 16.84 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'challenge_test_scramble' 500
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 1,497
'train' 67,389
'validation' 993
  • Estructura de características :
FeaturesDict({
    'concept_set_id': tf.int32,
    'concepts': Sequence(tf.string),
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'target': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
concepto_conjunto_id Tensor tf.int32
conceptos Secuencia (tensor) (Ninguna,) tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
  • Cita :
@inproceedings{lin2020commongen,
  title = "CommonGen: A Constrained Text Generation Challenge for Generative Commonsense Reasoning",
  author = "Lin, Bill Yuchen  and
    Zhou, Wangchunshu  and
    Shen, Ming  and
    Zhou, Pei  and
    Bhagavatula, Chandra  and
    Choi, Yejin  and
    Ren, Xiang",
  booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2020",
  month = nov,
  year = "2020",
  address = "Online",
  publisher = "Association for Computational Linguistics",
  url = "https://www.aclweb.org/anthology/2020.findings-emnlp.165",
  pages = "1823--1840",
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/cs_restaurantes

  • Descripción de la configuración : la tarea es generar respuestas en el contexto de un sistema de diálogo (hipotético) que proporciona información sobre restaurantes. La entrada es un tipo de acto de intención/diálogo básico y una lista de espacios (atributos) y sus valores. El resultado es una oración en lenguaje natural.

  • Tamaño de la descarga : 1.46 MiB

  • Tamaño del conjunto de datos : 2.71 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'challenge_test_scramble' 500
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 842
'train' 3,569
'validation' 781
  • Estructura de características :
FeaturesDict({
    'dialog_act': tf.string,
    'dialog_act_delexicalized': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'target': tf.string,
    'target_delexicalized': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
dialog_act Tensor tf.cadena
dialog_act_delexicalized Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
objetivo_delexicalizado Tensor tf.cadena
  • Cita :
@inproceedings{cs_restaurants,
  address = {Tokyo, Japan},
  title = {Neural {Generation} for {Czech}: {Data} and {Baselines}},
  shorttitle = {Neural {Generation} for {Czech}},
  url = {https://www.aclweb.org/anthology/W19-8670/},
  urldate = {2019-10-18},
  booktitle = {Proceedings of the 12th {International} {Conference} on {Natural} {Language} {Generation} ({INLG} 2019)},
  author = {Dušek, Ondřej and Jurčíček, Filip},
  month = oct,
  year = {2019},
  pages = {563--574}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/dardo

  • Descripción de la configuración : DART es un corpus de generación de registro de datos a texto estructurado de dominio abierto grande con anotaciones de oraciones de alta calidad en las que cada entrada es un conjunto de tripletas entidad-relación que siguen una ontología estructurada en árbol.

  • Tamaño de la descarga : 28.01 MiB

  • Tamaño del conjunto de datos : 33.78 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 6,959
'train' 62,659
'validation' 2,768
  • Estructura de características :
FeaturesDict({
    'dart_id': tf.int32,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'subtree_was_extended': tf.bool,
    'target': tf.string,
    'target_sources': Sequence(tf.string),
    'tripleset': Sequence(tf.string),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
dardo_id Tensor tf.int32
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
subtree_was_extended Tensor tf.bool
objetivo Tensor tf.cadena
fuentes_objetivo Secuencia (tensor) (Ninguna,) tf.cadena
conjunto triple Secuencia (tensor) (Ninguna,) tf.cadena
  • Cita :
@article{radev2020dart,
  title=Dart: Open-domain structured data record to text generation,
  author={Radev, Dragomir and Zhang, Rui and Rau, Amrit and Sivaprasad, Abhinand and Hsieh, Chiachun and Rajani, Nazneen Fatema and Tang, Xiangru and Vyas, Aadit and Verma, Neha and Krishna, Pranav and others},
  journal={arXiv preprint arXiv:2007.02871},
  year={2020}
}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/e2e_nlg

  • Descripción de configuración : el conjunto de datos E2E está diseñado para una tarea de datos a texto de dominio limitado: generación de descripciones/recomendaciones de restaurantes basadas en hasta 8 atributos diferentes (nombre, área, rango de precios, etc.)

  • Tamaño de descarga : 13.99 MiB

  • Tamaño del conjunto de datos : 16.92 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'challenge_test_scramble' 500
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 4,693
'train' 33,525
'validation' 4,299
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'meaning_representation': tf.string,
    'references': Sequence(tf.string),
    'target': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
significado_representacion Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
  • Cita :
@inproceedings{e2e_cleaned,
  address = {Tokyo, Japan},
  title = {Semantic {Noise} {Matters} for {Neural} {Natural} {Language} {Generation}},
  url = {https://www.aclweb.org/anthology/W19-8652/},
  booktitle = {Proceedings of the 12th {International} {Conference} on {Natural} {Language} {Generation} ({INLG} 2019)},
  author = {Dušek, Ondřej and Howcroft, David M and Rieser, Verena},
  year = {2019},
  pages = {421--426},
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/mlsum_de

  • Descripción de la configuración : MLSum es un conjunto de datos de resumen multilingüe a gran escala. Se construye a partir de medios de comunicación en línea, esta división se centra en el alemán.

  • Tamaño de la descarga : 345.98 MiB

  • Tamaño del conjunto de datos : 963.60 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'challenge_test_covid' 5,058
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 10,695
'train' 220,748
'validation' 11,392
  • Estructura de características :
FeaturesDict({
    'date': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'target': tf.string,
    'text': tf.string,
    'title': tf.string,
    'topic': tf.string,
    'url': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
fecha Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
texto Tensor tf.cadena
título Tensor tf.cadena
tema Tensor tf.cadena
URL Tensor tf.cadena
  • Cita :
@inproceedings{scialom-etal-2020-mlsum,
    title = "{MLSUM}: The Multilingual Summarization Corpus",
    author = {Scialom, Thomas  and Dray, Paul-Alexis  and Lamprier, Sylvain  and Piwowarski, Benjamin  and Staiano, Jacopo},
    booktitle = {Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
    year = {2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/mlsum_es

  • Descripción de la configuración : MLSum es un conjunto de datos de resumen multilingüe a gran escala. Se construye a partir de medios de noticias en línea, esta división se centra en el español.

  • Tamaño de la descarga : 501.27 MiB

  • Tamaño del conjunto de datos : 1.29 GiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'challenge_test_covid' 1,938
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 13,366
'train' 259,888
'validation' 9,977
  • Estructura de características :
FeaturesDict({
    'date': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'target': tf.string,
    'text': tf.string,
    'title': tf.string,
    'topic': tf.string,
    'url': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
fecha Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
texto Tensor tf.cadena
título Tensor tf.cadena
tema Tensor tf.cadena
URL Tensor tf.cadena
  • Cita :
@inproceedings{scialom-etal-2020-mlsum,
    title = "{MLSUM}: The Multilingual Summarization Corpus",
    author = {Scialom, Thomas  and Dray, Paul-Alexis  and Lamprier, Sylvain  and Piwowarski, Benjamin  and Staiano, Jacopo},
    booktitle = {Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
    year = {2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/schema_guided_dialog

  • Descripción de la configuración : el conjunto de datos del diálogo guiado por esquemas (SGD) contiene diálogos orientados a tareas multidominio de 18 000 entre un ser humano y un asistente virtual, que cubre 17 dominios que van desde bancos y eventos hasta medios, calendario, viajes y clima.

  • Tamaño de descarga : 17.00 MiB

  • Tamaño del conjunto de datos : 201.19 MiB

  • Auto-caché ( documentación ): Sí (challenge_test_backtranslation, challenge_test_bfp02, challenge_test_bfp05, challenge_test_nopunc, challenge_test_scramble, challenge_train_sample, challenge_validation_sample, test, validation), solo cuando shuffle_files=False (tren)

  • Divisiones :

Separar Ejemplos
'challenge_test_backtranslation' 500
'challenge_test_bfp02' 500
'challenge_test_bfp05' 500
'challenge_test_nopunc' 500
'challenge_test_scramble' 500
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 10,000
'train' 164,982
'validation' 10,000
  • Estructura de características :
FeaturesDict({
    'context': Sequence(tf.string),
    'dialog_acts': Sequence({
        'act': ClassLabel(shape=(), dtype=tf.int64, num_classes=18),
        'slot': tf.string,
        'values': Sequence(tf.string),
    }),
    'dialog_id': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'prompt': tf.string,
    'references': Sequence(tf.string),
    'service': tf.string,
    'target': tf.string,
    'turn_id': tf.int32,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
contexto Secuencia (tensor) (Ninguna,) tf.cadena
dialog_acts Secuencia
dialog_acts/act Etiqueta de clase tf.int64
diálogo_acts/ranura Tensor tf.cadena
dialog_acts/valores Secuencia (tensor) (Ninguna,) tf.cadena
diálogo_id Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
inmediato Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
Servicio Tensor tf.cadena
objetivo Tensor tf.cadena
turno_id Tensor tf.int32
  • Cita :
@article{rastogi2019towards,
  title={Towards Scalable Multi-domain Conversational Agents: The Schema-Guided Dialogue Dataset},
  author={Rastogi, Abhinav and Zang, Xiaoxue and Sunkara, Srinivas and Gupta, Raghav and Khaitan, Pranav},
  journal={arXiv preprint arXiv:1909.05855},
  year={2019}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/toto

  • Descripción de la configuración : ToTTo es una tarea NLG de tabla a texto. La tarea es la siguiente: dada una tabla de Wikipedia con nombres de filas, nombres de columnas y celdas de tabla, con un subconjunto de celdas resaltadas, generar una descripción en lenguaje natural para la parte resaltada de la tabla.

  • Tamaño de la descarga : 180.75 MiB

  • Tamaño del conjunto de datos : 645.86 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'challenge_test_scramble' 500
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 7,700
'train' 121,153
'validation' 7,700
  • Estructura de características :
FeaturesDict({
    'example_id': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'highlighted_cells': Sequence(Sequence(tf.int32)),
    'overlap_subset': tf.string,
    'references': Sequence(tf.string),
    'sentence_annotations': Sequence({
        'final_sentence': tf.string,
        'original_sentence': tf.string,
        'sentence_after_ambiguity': tf.string,
        'sentence_after_deletion': tf.string,
    }),
    'table': Sequence(Sequence({
        'column_span': tf.int32,
        'is_header': tf.bool,
        'row_span': tf.int32,
        'value': tf.string,
    })),
    'table_page_title': tf.string,
    'table_section_text': tf.string,
    'table_section_title': tf.string,
    'table_webpage_url': tf.string,
    'target': tf.string,
    'totto_id': tf.int32,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
ejemplo_id Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
celdas_resaltadas Secuencia(Secuencia(Tensor)) (Ninguno Ninguno) tf.int32
superposición_subconjunto Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
anotaciones_de_frases Secuencia
anotaciones_oracion/oracion_final Tensor tf.cadena
anotaciones_oracion/oracion_original Tensor tf.cadena
anotaciones_oración/oración_después_de_ambigüedad Tensor tf.cadena
anotaciones_frases/frases_después_de_supresión Tensor tf.cadena
mesa Secuencia
tabla/columna_span Tensor tf.int32
tabla/es_encabezado Tensor tf.bool
tabla/fila_span Tensor tf.int32
tabla/valor Tensor tf.cadena
título_página_tabla Tensor tf.cadena
table_section_text Tensor tf.cadena
table_section_title Tensor tf.cadena
table_webpage_url Tensor tf.cadena
objetivo Tensor tf.cadena
totto_id Tensor tf.int32
  • Cita :
@inproceedings{parikh2020totto,
  title=ToTTo: A Controlled Table-To-Text Generation Dataset,
  author={Parikh, Ankur and Wang, Xuezhi and Gehrmann, Sebastian and Faruqui, Manaal and Dhingra, Bhuwan and Yang, Diyi and Das, Dipanjan},
  booktitle={Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)},
  pages={1173--1186},
  year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gem/web_nlg_es

  • Descripción de la configuración : WebNLG es un conjunto de datos bilingüe (inglés, ruso) de conjuntos triples paralelos de DBpedia y textos breves que cubren alrededor de 450 propiedades diferentes de DBpedia. Los datos de WebNLG se crearon originalmente para promover el desarrollo de verbalizadores RDF capaces de generar texto corto y manejar la microplanificación.

  • Tamaño de la descarga : 12.57 MiB

  • Tamaño del conjunto de datos : 19.91 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'challenge_test_numbers' 500
'challenge_test_scramble' 500
'challenge_train_sample' 502
'challenge_validation_sample' 499
'test' 1,779
'train' 35,426
'validation' 1,667
  • Estructura de características :
FeaturesDict({
    'category': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'input': Sequence(tf.string),
    'references': Sequence(tf.string),
    'target': tf.string,
    'webnlg_id': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
categoría Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
aporte Secuencia (tensor) (Ninguna,) tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
webnlg_id Tensor tf.cadena
  • Cita :
@inproceedings{gardent2017creating,
  author = "Gardent, Claire
    and Shimorina, Anastasia
    and Narayan, Shashi
    and Perez-Beltrachini, Laura",
  title = "Creating Training Corpora for NLG Micro-Planners",
  booktitle = "Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
  year = "2017",
  publisher = "Association for Computational Linguistics",
  pages = "179--188",
  location = "Vancouver, Canada",
  doi = "10.18653/v1/P17-1017",
  url = "http://www.aclweb.org/anthology/P17-1017"
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

joya/web_nlg_ru

  • Descripción de la configuración : WebNLG es un conjunto de datos bilingüe (inglés, ruso) de conjuntos triples paralelos de DBpedia y textos breves que cubren alrededor de 450 propiedades diferentes de DBpedia. Los datos de WebNLG se crearon originalmente para promover el desarrollo de verbalizadores RDF capaces de generar texto corto y manejar la microplanificación.

  • Tamaño de la descarga : 7.49 MiB

  • Tamaño del conjunto de datos : 11.30 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'challenge_test_scramble' 500
'challenge_train_sample' 501
'challenge_validation_sample' 500
'test' 1,102
'train' 14,630
'validation' 790
  • Estructura de características :
FeaturesDict({
    'category': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'input': Sequence(tf.string),
    'references': Sequence(tf.string),
    'target': tf.string,
    'webnlg_id': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
categoría Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
aporte Secuencia (tensor) (Ninguna,) tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
webnlg_id Tensor tf.cadena
  • Cita :
@inproceedings{gardent2017creating,
  author = "Gardent, Claire
    and Shimorina, Anastasia
    and Narayan, Shashi
    and Perez-Beltrachini, Laura",
  title = "Creating Training Corpora for NLG Micro-Planners",
  booktitle = "Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)",
  year = "2017",
  publisher = "Association for Computational Linguistics",
  pages = "179--188",
  location = "Vancouver, Canada",
  doi = "10.18653/v1/P17-1017",
  url = "http://www.aclweb.org/anthology/P17-1017"
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_auto_asset_turk

  • Descripción de la configuración : WikiAuto proporciona un conjunto de oraciones alineadas de Wikipedia en inglés y Wikipedia en inglés simple como recurso para entrenar sistemas de simplificación de oraciones. ASSET y TURK son conjuntos de datos de simplificación de alta calidad que se utilizan para las pruebas.

  • Tamaño de la descarga : 121.01 MiB

  • Tamaño del conjunto de datos : 202.40 MiB

  • Auto-cached ( documentation ): Yes (challenge_test_asset_backtranslation, challenge_test_asset_bfp02, challenge_test_asset_bfp05, challenge_test_asset_nopunc, challenge_test_turk_backtranslation, challenge_test_turk_bfp02, challenge_test_turk_bfp05, challenge_test_turk_nopunc, challenge_train_sample, challenge_validation_sample, test_asset, test_turk, validation), Only when shuffle_files=False (train)

  • Divisiones :

Separar Ejemplos
'challenge_test_asset_backtranslation' 359
'challenge_test_asset_bfp02' 359
'challenge_test_asset_bfp05' 359
'challenge_test_asset_nopunc' 359
'challenge_test_turk_backtranslation' 359
'challenge_test_turk_bfp02' 359
'challenge_test_turk_bfp05' 359
'challenge_test_turk_nopunc' 359
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test_asset' 359
'test_turk' 359
'train' 483,801
'validation' 20,000
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'target': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
objetivo Tensor tf.cadena
  • Cita :
@inproceedings{jiang-etal-2020-neural,
    title = "Neural {CRF} Model for Sentence Alignment in Text Simplification",
    author = "Jiang, Chao  and
      Maddela, Mounica  and
      Lan, Wuwei  and
      Zhong, Yang  and
      Xu, Wei",
    booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics",
    month = jul,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://www.aclweb.org/anthology/2020.acl-main.709",
    doi = "10.18653/v1/2020.acl-main.709",
    pages = "7943--7960",
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/xsum

  • Descripción de configuración : el conjunto de datos es para la tarea de resumen abstracto en su forma extrema, se trata de resumir un documento en una sola oración.

  • Tamaño de la descarga : 246.31 MiB

  • Tamaño del conjunto de datos : 78.89 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'challenge_test_backtranslation' 500
'challenge_test_bfp_02' 500
'challenge_test_bfp_05' 500
'challenge_test_covid' 401
'challenge_test_nopunc' 500
'challenge_train_sample' 500
'challenge_validation_sample' 500
'test' 1,166
'train' 23,206
'validation' 1,117
  • Estructura de características :
FeaturesDict({
    'document': tf.string,
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'target': tf.string,
    'xsum_id': tf.string,
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
documento Tensor tf.cadena
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
objetivo Tensor tf.cadena
xsum_id Tensor tf.cadena
  • Cita :
@inproceedings{Narayan2018dont,
  author = "Shashi Narayan and Shay B. Cohen and Mirella Lapata",
  title = "Don't Give Me the Details, Just the Summary! {T}opic-Aware Convolutional Neural Networks for Extreme Summarization",
  booktitle = "Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing ",
  year = "2018",
  address = "Brussels, Belgium",
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_arabic_ar

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de descarga : 56.25 MiB

  • Tamaño del conjunto de datos : 291.42 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 5,841
'train' 20,441
'validation' 2,919
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'ar': Text(shape=(), dtype=tf.string),
        'en': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'ar': Text(shape=(), dtype=tf.string),
        'en': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
fuente_alineada/ar Texto tf.cadena
source_aligned/es Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
objetivo_alineado/ar Texto tf.cadena
target_aligned/es Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_chinese_zh

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 31.38 MiB

  • Tamaño del conjunto de datos : 122.06 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 3,775
'train' 13,211
'validation' 1,886
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'zh': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'zh': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/zh Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/zh Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_czech_cs

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 13.84 MiB

  • Tamaño del conjunto de datos : 58.05 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 1,438
'train' 5,033
'validation' 718
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'cs': Text(shape=(), dtype=tf.string),
        'en': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'cs': Text(shape=(), dtype=tf.string),
        'en': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
fuente_alineada/cs Texto tf.cadena
source_aligned/es Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
objetivo_alineado/cs Texto tf.cadena
target_aligned/es Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_dutch_nl

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 53.88 MiB

  • Tamaño del conjunto de datos : 237.97 MiB

  • Almacenamiento automático en caché ( documentación ): Sí (prueba, validación), solo cuando shuffle_files=False (tren)

  • Divisiones :

Separar Ejemplos
'test' 6,248
'train' 21,866
'validation' 3,123
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'nl': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'nl': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/nl Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/nl Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_english_en

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 112.56 MiB

  • Tamaño del conjunto de datos : 657.51 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 28,614
'train' 99,020
'validation' 13,823
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_french_fr

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 113.26 MiB

  • Tamaño del conjunto de datos : 522.28 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 12,731
'train' 44,556
'validation' 6,364
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'fr': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'fr': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
alineado_fuente/fr Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
target_aligned/fr Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_german_de

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 102.65 MiB

  • Tamaño del conjunto de datos : 452.46 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 11,669
'train' 40,839
'validation' 5,833
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'de': Text(shape=(), dtype=tf.string),
        'en': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'de': Text(shape=(), dtype=tf.string),
        'en': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
alineado_fuente/de Texto tf.cadena
source_aligned/es Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
objetivo_alineado/de Texto tf.cadena
target_aligned/es Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_hindi_hi

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 20.07 MiB

  • Tamaño del conjunto de datos : 138.06 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 1,984
'train' 6,942
'validation' 991
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'hi': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'hi': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/hola Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/hola Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_indonesian_id

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de descarga : 80.08 MiB

  • Tamaño del conjunto de datos : 370.63 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 9,497
'train' 33,237
'validation' 4,747
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'id': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'id': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/id Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
target_aligned/id Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_italian_it

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 84.80 MiB

  • Tamaño del conjunto de datos : 374.40 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 10,189
'train' 35,661
'validation' 5,093
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'it': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'it': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/es Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/es Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_japanese_ja

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 21.75 MiB

  • Tamaño del conjunto de datos : 103.19 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 2,530
'train' 8,853
'validation' 1,264
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'ja': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'ja': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/ja Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/ja Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_korean_ko

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 22.26 MiB

  • Tamaño del conjunto de datos : 102.35 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 2,436
'train' 8,524
'validation' 1,216
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'ko': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'ko': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/ko Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/ko Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_portugués_pt

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 131.17 MiB

  • Tamaño del conjunto de datos : 570.46 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 16,331
'train' 57,159
'validation' 8,165
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'pt': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'pt': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/pt Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/pt Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_ruso_ru

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 101.36 MiB

  • Tamaño del conjunto de datos : 564.69 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 10,580
'train' 37,028
'validation' 5,288
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'ru': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'ru': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/ru Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/ru Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gem/wiki_lingua_spanish_es

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 189.06 MiB

  • Tamaño del conjunto de datos : 849.75 MiB

  • Almacenamiento automático en caché ( documentación ): No

  • Divisiones :

Separar Ejemplos
'test' 22,632
'train' 79,212
'validation' 11,316
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'es': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'es': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/es Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/es Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_thai_th

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de descarga : 28.60 MiB

  • Tamaño del conjunto de datos : 193.77 MiB

  • Almacenamiento automático en caché ( documentación ): Sí (prueba, validación), solo cuando shuffle_files=False (tren)

  • Divisiones :

Separar Ejemplos
'test' 2,950
'train' 10,325
'validation' 1,475
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'th': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'th': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/th Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/th Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_turkish_tr

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de la descarga : 6.73 MiB

  • Tamaño del conjunto de datos : 30.75 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 900
'train' 3,148
'validation' 449
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'tr': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'tr': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/tr Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/tr Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."

gema/wiki_lingua_vietnamita_vi

  • Descripción de la configuración : Wikilingua es un conjunto de datos multilingüe a gran escala para la evaluación de sistemas de resumen abstracto multilingüe.

  • Tamaño de descarga : 36.27 MiB

  • Tamaño del conjunto de datos : 179.77 MiB

  • Almacenamiento automático en caché ( documentación ): Sí

  • Divisiones :

Separar Ejemplos
'test' 3,917
'train' 13,707
'validation' 1,957
  • Estructura de características :
FeaturesDict({
    'gem_id': tf.string,
    'gem_parent_id': tf.string,
    'references': Sequence(tf.string),
    'source': tf.string,
    'source_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'vi': Text(shape=(), dtype=tf.string),
    }),
    'target': tf.string,
    'target_aligned': Translation({
        'en': Text(shape=(), dtype=tf.string),
        'vi': Text(shape=(), dtype=tf.string),
    }),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
gem_id Tensor tf.cadena
gem_parent_id Tensor tf.cadena
referencias Secuencia (tensor) (Ninguna,) tf.cadena
fuente Tensor tf.cadena
fuente_alineada Traducción
source_aligned/es Texto tf.cadena
fuente_alineada/vi Texto tf.cadena
objetivo Tensor tf.cadena
objetivo_alineado Traducción
target_aligned/es Texto tf.cadena
objetivo_alineado/vi Texto tf.cadena
  • Cita :
@inproceedings{ladhak-wiki-2020,
title=WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization,
author={Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},
booktitle={Findings of EMNLP, 2020},
year={2020}
}
@article{gehrmann2021gem,
  author    = {Sebastian Gehrmann and
               Tosin P. Adewumi and
               Karmanya Aggarwal and
               Pawan Sasanka Ammanamanchi and
               Aremu Anuoluwapo and
               Antoine Bosselut and
               Khyathi Raghavi Chandu and
               Miruna{-}Adriana Clinciu and
               Dipanjan Das and
               Kaustubh D. Dhole and
               Wanyu Du and
               Esin Durmus and
               Ondrej Dusek and
               Chris Emezue and
               Varun Gangal and
               Cristina Garbacea and
               Tatsunori Hashimoto and
               Yufang Hou and
               Yacine Jernite and
               Harsh Jhamtani and
               Yangfeng Ji and
               Shailza Jolly and
               Dhruv Kumar and
               Faisal Ladhak and
               Aman Madaan and
               Mounica Maddela and
               Khyati Mahajan and
               Saad Mahamood and
               Bodhisattwa Prasad Majumder and
               Pedro Henrique Martins and
               Angelina McMillan{-}Major and
               Simon Mille and
               Emiel van Miltenburg and
               Moin Nadeem and
               Shashi Narayan and
               Vitaly Nikolaev and
               Rubungo Andre Niyongabo and
               Salomey Osei and
               Ankur P. Parikh and
               Laura Perez{-}Beltrachini and
               Niranjan Ramesh Rao and
               Vikas Raunak and
               Juan Diego Rodriguez and
               Sashank Santhanam and
               Jo{\~{a}}o Sedoc and
               Thibault Sellam and
               Samira Shaikh and
               Anastasia Shimorina and
               Marco Antonio Sobrevilla Cabezudo and
               Hendrik Strobelt and
               Nishant Subramani and
               Wei Xu and
               Diyi Yang and
               Akhila Yerukola and
               Jiawei Zhou},
  title     = {The {GEM} Benchmark: Natural Language Generation, its Evaluation and
               Metrics},
  journal   = {CoRR},
  volume    = {abs/2102.01672},
  year      = {2021},
  url       = {https://arxiv.org/abs/2102.01672},
  archivePrefix = {arXiv},
  eprint    = {2102.01672}
}

Note that each GEM dataset has its own citation. Please see the source to see
the correct citation for each contained dataset."