/RussianSuperGLUE

Russian SuperGLUE benchmark

Primary LanguageJupyter NotebookMIT LicenseMIT

RussianSuperGLUE

Russian SuperGLUE benchmark

We introduce an advanced Russian general language understanding evaluation benchmark.

Recent advances in the field of universal language models and transformers require the development of a methodology for their broad diagnostics and testing for general intellectual skills - detection of natural language inference, commonsense reasoning, ability to perform simple logical operations regardless of text subject or lexicon. For the first time, a benchmark of nine tasks, collected and organized analogically to the SuperGLUE methodology, was developed from scratch for the Russian language. We provide baselines, human level evaluation, an open-source framework for evaluating models and an overall leaderboard of transformer models for the Russian language.

RELEASE v1.1

  • update and expand some datasets:
    • RUSSE: new test + human benchmark
    • DaNetQA: expand the dataset + new test + human benchmark
    • RuCoS: expand the dataset + clean typos/inaccuracies
    • MuSeRC: expand the dataset + clean typos/inaccuracies
  • add and improve code for jiant:
    • evaluation of models: GPT-3, GPT-2
    • correct lidirus preprocessing
  • fix typos and bugs
  • refactor web interface and improved reliability of the model evaluation system on the website

Instructions:

Jupyter link

Leaderboard:

Russiansuperglue.com

Download the Data:

All the tasks (zip)

Some tasks from the website

Documentation:

You can see our documentation at diagnostics description

Cite us:

Read our article

Please, cite us this way:

@inproceedings{shavrina-etal-2020-russiansuperglue,
    title = "{R}ussian{S}uper{GLUE}: A {R}ussian Language Understanding Evaluation Benchmark",
    author = "Shavrina, Tatiana  and
      Fenogenova, Alena  and
      Anton, Emelyanov  and
      Shevelev, Denis  and
      Artemova, Ekaterina  and
      Malykh, Valentin  and
      Mikhailov, Vladislav  and
      Tikhonova, Maria  and
      Chertok, Andrey  and
      Evlampiev, Andrey",
    booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)",
    month = nov,
    year = "2020",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2020.emnlp-main.381",
    doi = "10.18653/v1/2020.emnlp-main.381",
    pages = "4717--4726",
    abstract = "In this paper, we introduce an advanced Russian general language understanding evaluation benchmark {--} Russian SuperGLUE. Recent advances in the field of universal language models and transformers require the development of a methodology for their broad diagnostics and testing for general intellectual skills - detection of natural language inference, commonsense reasoning, ability to perform simple logical operations regardless of text subject or lexicon. For the first time, a benchmark of nine tasks, collected and organized analogically to the SuperGLUE methodology, was developed from scratch for the Russian language. We also provide baselines, human level evaluation, open-source framework for evaluating models, and an overall leaderboard of transformer models for the Russian language. Besides, we present the first results of comparing multilingual models in the translated diagnostic test set and offer the first steps to further expanding or assessing State-of-the-art models independently of language.",
}