SEACrowd v0.2.0 via pip package

Find seacrowd library (v0.2.0) at https://pypi.org/project/seacrowd/.

Installation

pip install seacrowd

Ready-for-use methods

• list_datasets()
• load_dataset(...)
• load_datasets(...)
• load_dataset_by_config_name(...)
• load_datasets_by_config_names(...)
• for_dataset(...)
• for_datasets(...)
• for_config_name(...)
• for_config_names(...)
• list_benchmarks()
• load_benchmark(...)
• available_dataset_names(...)
• available_config_names(...)

Other methods haven't been tested and verified.

---

If you are using any resources from SEACrowd, including datasheets, dataloaders, code, etc., please cite the following publication:

@article{lovenia2024seacrowd,
      title={SEACrowd: A Multilingual Multimodal Data Hub and Benchmark Suite for Southeast Asian Languages}, 
      author={Holy Lovenia and Rahmad Mahendra and Salsabil Maulana Akbar and Lester James V. Miranda and Jennifer Santoso and Elyanah Aco and Akhdan Fadhilah and Jonibek Mansurov and Joseph Marvin Imperial and Onno P. Kampman and Joel Ruben Antony Moniz and Muhammad Ravi Shulthan Habibi and Frederikus Hudi and Railey Montalan and Ryan Ignatius and Joanito Agili Lopo and William Nixon and Börje F. Karlsson and James Jaya and Ryandito Diandaru and Yuze Gao and Patrick Amadeus and Bin Wang and Jan Christian Blaise Cruz and Chenxi Whitehouse and Ivan Halim Parmonangan and Maria Khelli and Wenyu Zhang and Lucky Susanto and Reynard Adha Ryanda and Sonny Lazuardi Hermawan and Dan John Velasco and Muhammad Dehan Al Kautsar and Willy Fitra Hendria and Yasmin Moslem and Noah Flynn and Muhammad Farid Adilazuarda and Haochen Li and Johanes Lee and R. Damanhuri and Shuo Sun and Muhammad Reza Qorib and Amirbek Djanibekov and Wei Qi Leong and Quyet V. Do and Niklas Muennighoff and Tanrada Pansuwan and Ilham Firdausi Putra and Yan Xu and Ngee Chia Tai and Ayu Purwarianti and Sebastian Ruder and William Tjhi and Peerat Limkonchotiwat and Alham Fikri Aji and Sedrick Keh and Genta Indra Winata and Ruochen Zhang and Fajri Koto and Zheng-Xin Yong and Samuel Cahyawijaya},
      year={2024},
      eprint={2406.10118},
      journal={arXiv preprint arXiv: 2406.10118}
}

SEACrowd v0.1.3 via pip package

Find seacrowd library (v0.1.3) at https://pypi.org/project/seacrowd/.

Installation

pip install seacrowd

Ready-for-use methods

• list_datasets()
• load_dataset(...)
• load_datasets(...)
• load_dataset_by_config_name(...)
• load_datasets_by_config_names(...)
• for_dataset(...)
• for_datasets(...)
• for_config_name(...)
• for_config_names(...)
• list_benchmarks()
• load_benchmark(...)

Other methods haven't been tested and verified.

Simple usage examples

import pprint

import seacrowd as sc

if __name__ == "__main__":
    print('=== LIST DATASETS ===')
    dset_names = sc.list_datasets()
    pprint.pprint(dset_names[:10])
    print()

    print('=== LIST DATASETS (with config names) ===')
    dset_names = sc.list_datasets(with_config=True)
    pprint.pprint({k: dset_names[k] for k in list(dset_names)[:10]})
    print()

    print('=== LOAD DATASET ===')
    dset = sc.load_dataset(dataset_name='khpos', schema='seacrowd')
    print(dset)
    print()

    print('=== LOAD DATASETS ===')
    dsets = sc.load_datasets(
        dataset_names=["thai_sum", "vsolscsum"], schema='seacrowd_t2t')
    pprint.pprint(dsets)
    print()

    print(f'=== LIST BENCHMARKS ===')
    benchmark_names = sc.list_benchmarks()
    pprint.pprint(benchmark_names)
    print()

    print(f'=== LOAD BENCHMARK ===')
    benchmark_dsets = sc.load_benchmark("SEACrowd-VL")
    pprint.pprint(benchmark_dsets)
    print()

Output

100%|██████████████████████████████████████████████████████████████████████████████| 406/406 [00:51<00:00,  7.81it/s]
=== LIST DATASETS ===
['abui_wordnet',
 'ac_iquad',
 'alice_thi',
 'alorese',
 'alt_burmese_treebank',
 'ara_close',
 'asr_ibsc',
 'asr_indocsc',
 'asr_malcsc',
 'asr_sindodusc']

=== LIST DATASETS (with config names) ===
{'abui_wordnet': ['abui_wordnet_source'],
 'ac_iquad': ['ac_iquad_simple_source',
              'ac_iquad_simple_seacrowd_qa',
              'ac_iquad_complex_source',
              'ac_iquad_complex_seacrowd_qa'],
 'alice_thi': ['alice_thi_THI-D10_source',
               'alice_thi_THI-D10_seacrowd_imtext',
               'alice_thi_THI-C68_source',
               'alice_thi_THI-C68_seacrowd_imtext'],
 'alorese': ['alorese_source',
             'alorese_seacrowd_t2t',
             'alorese_seacrowd_sptext'],
 'alt_burmese_treebank': ['alt_burmese_treebank_source',
                          'alt_burmese_treebank_seacrowd_tree'],
 'ara_close': ['ara_close_bcl_source',
               'ara_close_ceb_source',
               'ara_close_bcl_seacrowd_text',
               'ara_close_ceb_seacrowd_text'],
 'asr_ibsc': ['asr_ibsc_source', 'asr_ibsc_seacrowd_sptext'],
 'asr_indocsc': ['asr_indocsc_source', 'asr_indocsc_seacrowd_sptext'],
 'asr_malcsc': ['asr_malcsc_source', 'asr_malcsc_seacrowd_sptext'],
 'asr_sindodusc': ['asr_sindodusc_source', 'asr_sindodusc_seacrowd_sptext']}

=== LOAD DATASET ===
Downloading data: 100%|█████████████████████████████████████████████████████████| 2.37M/2.37M [00:00<00:00, 71.5MB/s]
Downloading data: 197kB [00:00, 61.5MB/s]
Downloading data: 191kB [00:00, 62.3MB/s]
Generating train split: 12000 examples [00:01, 11332.38 examples/s]
Generating validation split: 1000 examples [00:00, 9529.80 examples/s]
Generating test split: 1000 examples [00:00, 8603.38 examples/s]
DatasetDict({
    train: Dataset({
        features: ['id', 'tokens', 'labels'],
        num_rows: 12000
    })
    validation: Dataset({
        features: ['id', 'tokens', 'labels'],
        num_rows: 1000
    })
    test: Dataset({
        features: ['id', 'tokens', 'labels'],
        num_rows: 1000
    })
})

=== LOAD DATASETS ===
Downloading data: 100%|█████████████████████████████████████████████████████████| 2.94G/2.94G [02:47<00:00, 17.5MB/s]
Downloading data: 100%|███████████████████████████████████████████████████████████| 118M/118M [00:07<00:00, 16.2MB/s]
Downloading data: 100%|███████████████████████████████████████████████████████████| 119M/119M [00:07<00:00, 15.8MB/s]
Generating train split: 358868 examples [01:19, 4528.13 examples/s]
Generating test split: 11000 examples [00:03, 3633.71 examples/s]
Generating validation split: 11000 examples [00:02, 3700.06 examples/s]
Downloading data: 100%|█████████████████████████████████████████████████████████| 3.20M/3.20M [00:00<00:00, 57.1MB/s]
Generating train split: 141 examples [00:00, 416.90 examples/s]
[DatasetDict({
    train: Dataset({
        features: ['id', 'text_1', 'text_2', 'text_1_name', 'text_2_name'],
        num_rows: 358868
    })
    test: Dataset({
        features: ['id', 'text_1', 'text_2', 'text_1_name', 'text_2_name'],
        num_rows: 11000
    })
    validation: Dataset({
        features: ['id', 'text_1', 'text_2', 'text_1_name', 'text_2_name'],
        num_rows: 11000
    })
}),
DatasetDict({
    train: Dataset({
        features: ['id', 'text_1', 'text_2', 'text_1_name', 'text_2_name'],
        num_rows: 141
    })
})]

=== LIST BENCHMARKS ===
['IndoNLU',
 'IndoNLG',
 'IndoLEM',
 'NusaX',
 'NusaNLU',
 'NusaNLG',
 'NusaASR',
 'NusaTranslation',
 'NusaParagraph',
 'NusaWrites',
 'SEACrowd-NLU',
 'SEACrowd-NLG',
 'SEACrowd-Speech',
 'SEACrowd-VL']

=== LOAD BENCHMARK ===
Downloading data: 100%|██████████████████████████████████████████████████████████| 16.8M/16.8M [00:00<00:00, 100MB/s]
Downloading data:  87%|███████████████████████████████████████████████████        | 274M/316M [00:33<00:03, 13.7MB/s]
Downloading data:  87%|███████████████████████████████████████████████████▎       | 275M/316M [00:33<00:03, 13.5MB/s]
Downloading data:  87%|███████████████████████████████████████████████████▌       | 277M/316M [00:33<00:02, 13.3MB/s]
Downloading data:  88%|███████████████████████████████████████████████████▊       | 278M/316M [00:33<00:02, 13.1MB/s]
Downloading data: 100%|███████████████████████████████████████████████████████████| 316M/316M [00:36<00:00, 8.75MB/s]
Downloading data: 1.30MB [00:00, 57.2MB/s]
Generating train split: 3449 examples [00:00, 22590.52 examples/s]
Generating test split: 2760 examples [00:00, 24525.70 examples/s]
Generating validation split: 900 examples [00:00, 20227.38 examples/s]
Generating train split: 3447 examples [00:00, 22407.32 examples/s]
Generating test split: 2775 examples [00:00, 24173.28 examples/s]
Generating validation split: 904 examples [00:00, 19832.99 examples/s]
Generating train split: 3490 examples [00:00, 21186.11 examples/s]
Generating test split: 2798 examples [00:00, 23490.40 examples/s]
Generating validation split: 912 examples [00:00, 20180.77 examples/s]
Generating train split: 3557 examples [00:00, 22074.76 examples/s]
Generating test split: 2855 examples [00:00, 23756.76 examples/s]
Generating validation split: 938 examples [00:00, 19680.73 examples/s]
[DatasetDict({
    train: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 3449
    })
    test: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 2760
    })
    validation: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 900
    })
}),
 DatasetDict({
    train: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 3447
    })
    test: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 2775
    })
    validation: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 904
    })
}),
 DatasetDict({
    train: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 3490
    })
    test: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 2798
    })
    validation: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 912
    })
}),
 DatasetDict({
    train: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 3557
    })
    test: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 2855
    })
    validation: Dataset({
        features: ['id', 'image_paths', 'texts', 'metadata'],
        num_rows: 938
    })
})]

---

If you are using any resources from SEACrowd, including datasheets, dataloaders, code, etc., please cite the following publication:

@article{lovenia2024seacrowd,
      title={SEACrowd: A Multilingual Multimodal Data Hub and Benchmark Suite for Southeast Asian Languages}, 
      author={Holy Lovenia and Rahmad Mahendra and Salsabil Maulana Akbar and Lester James V. Miranda and Jennifer Santoso and Elyanah Aco and Akhdan Fadhilah and Jonibek Mansurov and Joseph Marvin Imperial and Onno P. Kampman and Joel Ruben Antony Moniz and Muhammad Ravi Shulthan Habibi and Frederikus Hudi and Railey Montalan and Ryan Ignatius and Joanito Agili Lopo and William Nixon and Börje F. Karlsson and James Jaya and Ryandito Diandaru and Yuze Gao and Patrick Amadeus and Bin Wang and Jan Christian Blaise Cruz and Chenxi Whitehouse and Ivan Halim Parmonangan and Maria Khelli and Wenyu Zhang and Lucky Susanto and Reynard Adha Ryanda and Sonny Lazuardi Hermawan and Dan John Velasco and Muhammad Dehan Al Kautsar and Willy Fitra Hendria and Yasmin Moslem and Noah Flynn and Muhammad Farid Adilazuarda and Haochen Li and Johanes Lee and R. Damanhuri and Shuo Sun and Muhammad Reza Qorib and Amirbek Djanibekov and Wei Qi Leong and Quyet V. Do and Niklas Muennighoff and Tanrada Pansuwan and Ilham Firdausi Putra and Yan Xu and Ngee Chia Tai and Ayu Purwarianti and Sebastian Ruder and William Tjhi and Peerat Limkonchotiwat and Alham Fikri Aji a...