Datasets:

Modalities:
Tabular
Text
Formats:
parquet
ArXiv:
Libraries:
Datasets
Dask
License:

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Aya_Kannada

This Dataset is curated from the original Aya-Collection dataset that was open-sourced by Cohere under the Apache-2.0 license. The Aya Collection is a massive multilingual collection comprising 513 million instances of prompts and completions that cover a wide range of tasks. This collection uses instruction-style templates from fluent speakers and applies them to a curated list of datasets. It also includes translations of instruction-style datasets into 101 languages. The Aya Dataset, a human-curated multilingual instruction and response dataset, is part of this collection. Refer to our paper for more details about the collection.

Motivations & Intentions

The original dataset is large and more task-specific than language-specific. To carry out a task specific to the Indic language, one would previously have needed to download the entire dataset (~600 GB) and filter it. As we were training an Indic LLm internally, we filtered the dataset by language and curated this dataset. You can find all the Indic-language specific datasets - here.

Data Instances

An example of a train instance looks as follows:

{'id': 246001,
 'inputs': 'The following query in English is taken from the geography category. What could be the answer to the question?\nWhat is the seventh tallest mountain in North America?',
 'targets': 'The answer is Mount Lucania.',
 'dataset_name': 'Mintaka-inst',
 'sub_dataset_name': '-',
 'task_type': 'question-answering',
 'template_id': 3,
 'language': 'eng',
 'split': 'train',
 'script': 'Latn'
}

Data Fields

The data fields are the same among all splits:

  • id: Unique id of the data point
  • inputs: Prompt or input to the language model.
  • targets: Completion or output of the language model.
  • dataset_name: The name of the source dataset that the data point was taken from
  • sub_dataset_name: If the source is a collection, this field indicates which part of that collection the data point was taken from. If it is not a collection, this field is left blank.
  • task_type: The task type that this conversation belongs to.
  • template_id: The id of the template applied to this data point.
  • language: The ISO code of the dialect of the conversation.
  • script: The script of the language.
  • split: Indicates whether the data point is part of the train or the test split.

Licensing Information

This dataset can be used for any purpose, whether academic or commercial, under the terms of the Apache 2.0 License.

Citation

@misc{singh2024aya,
      title={Aya Dataset: An Open-Access Collection for Multilingual Instruction Tuning}, 
      author={Shivalika Singh and Freddie Vargus and Daniel Dsouza and Börje F. Karlsson and Abinaya Mahendiran and Wei-Yin Ko and Herumb Shandilya and Jay Patel and Deividas Mataciunas and Laura OMahony and Mike Zhang and Ramith Hettiarachchi and Joseph Wilson and Marina Machado and Luisa Souza Moura and Dominik Krzemiński and Hakimeh Fadaei and Irem Ergün and Ifeoma Okoh and Aisha Alaagib and Oshan Mudannayake and Zaid Alyafeai and Vu Minh Chien and Sebastian Ruder and Surya Guthikonda and Emad A. Alghamdi and Sebastian Gehrmann and Niklas Muennighoff and Max Bartolo and Julia Kreutzer and Ahmet Üstün and Marzieh Fadaee and Sara Hooker},
      year={2024},
      eprint={2402.06619},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}
Downloads last month
109

Collection including Cognitive-Lab/Aya_Kannada