diff --git "a/dataset_infos.json" "b/dataset_infos.json" --- "a/dataset_infos.json" +++ "b/dataset_infos.json" @@ -1 +1 @@ -{"cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 15313824, "num_examples": 5033, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 2283890, "num_examples": 718, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 4412468, "num_examples": 1438, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 2295378, "num_examples": 718, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 4435476, "num_examples": 1438, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 28741036, "size_in_bytes": 2404231579}, "hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 47445942, "num_examples": 6942, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6883899, "num_examples": 991, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13405183, "num_examples": 1984, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6899755, "num_examples": 991, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13436927, "num_examples": 1984, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 88071706, "size_in_bytes": 2463562249}, "ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 29890060, "num_examples": 8853, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4268616, "num_examples": 1264, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 8380066, "num_examples": 2530, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4288840, "num_examples": 1264, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8418996, "num_examples": 2530, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 55246578, "size_in_bytes": 2430737121}, "zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 30670554, "num_examples": 13211, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4482747, "num_examples": 1886, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 8734304, "num_examples": 3775, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4512923, "num_examples": 1886, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 6948840, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 55349368, "size_in_bytes": 2430839911}, "ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 28479918, "num_examples": 8524, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4103150, "num_examples": 1216, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 8053102, "num_examples": 2436, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4122606, "num_examples": 1216, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8092078, "num_examples": 2436, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 52850854, "size_in_bytes": 2428341397}, "th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 63845006, "num_examples": 10325, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 9514117, "num_examples": 1475, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 18065846, "num_examples": 2950, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9537717, "num_examples": 1475, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 18113046, "num_examples": 2950, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 119075732, "size_in_bytes": 2494566275}, "fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 145666997, "num_examples": 44556, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21014635, "num_examples": 6364, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 41609316, "num_examples": 12731, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9941910, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 10002143, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 228235001, "size_in_bytes": 2603725544}, "nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 64659473, "num_examples": 21866, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 9485039, "num_examples": 3123, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 18372924, "num_examples": 6248, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9119620, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8878901, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 110515957, "size_in_bytes": 2486006500}, "tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 8628113, "num_examples": 3148, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 1226412, "num_examples": 449, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 2358687, "num_examples": 900, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 1233596, "num_examples": 449, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 2373087, "num_examples": 900, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 15819895, "size_in_bytes": 2391310438}, "vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 51583452, "num_examples": 13707, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7655401, "num_examples": 1957, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 14780646, "num_examples": 3917, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7686713, "num_examples": 1957, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 11411460, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 93117672, "size_in_bytes": 2468608215}, "it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 103856992, "num_examples": 35661, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 14884822, "num_examples": 5093, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29230180, "num_examples": 10189, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8815371, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8597856, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 165385221, "size_in_bytes": 2540875764}, "ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 180589541, "num_examples": 37028, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26071057, "num_examples": 5288, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 50905392, "num_examples": 10580, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14886738, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14442645, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 286895373, "size_in_bytes": 2662385916}, "de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 126056160, "num_examples": 40839, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18178743, "num_examples": 5833, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 35827004, "num_examples": 11669, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9489219, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 9280467, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 198831593, "size_in_bytes": 2574322136}, "id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 101649811, "num_examples": 33237, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 14732392, "num_examples": 4747, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 28785267, "num_examples": 9497, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9390452, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 9092274, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 163650196, "size_in_bytes": 2539140739}, "ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 87641376, "num_examples": 20441, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12819112, "num_examples": 2919, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 25185873, "num_examples": 5841, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12865816, "num_examples": 2919, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13142193, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 151654370, "size_in_bytes": 2527144913}, "pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 153451688, "num_examples": 57159, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22004444, "num_examples": 8165, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 43442458, "num_examples": 16331, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8165197, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7865967, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 234929754, "size_in_bytes": 2610420297}, "en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 257362818, "num_examples": 99020, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35939502, "num_examples": 13823, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73923383, "num_examples": 28614, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7893608, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7709689, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 382829000, "size_in_bytes": 2758319543}, "es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 235253742, "num_examples": 79212, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33645613, "num_examples": 11316, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67245976, "num_examples": 22632, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8990297, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8931532, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 354067160, "size_in_bytes": 2729557703}, "cs_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 83267533, "num_examples": 16019, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11826763, "num_examples": 2201, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23175474, "num_examples": 4468, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11861979, "num_examples": 2201, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23246962, "num_examples": 4468, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 153378711, "size_in_bytes": 2528869254}, "cs_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 58487808, "num_examples": 17942, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8444442, "num_examples": 2518, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16171454, "num_examples": 4942, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8484730, "num_examples": 2518, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16248976, "num_examples": 4942, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 107837410, "size_in_bytes": 2483327953}, "cs_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 61973073, "num_examples": 24110, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8973177, "num_examples": 3376, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 17408183, "num_examples": 6749, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9027193, "num_examples": 3376, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15670303, "num_examples": 5974, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 113051929, "size_in_bytes": 2488542472}, "cs_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 59976457, "num_examples": 18463, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8491961, "num_examples": 2566, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16518964, "num_examples": 5074, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8533017, "num_examples": 2566, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16600148, "num_examples": 5074, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 110120547, "size_in_bytes": 2485611090}, "cs_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 108360356, "num_examples": 21378, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15820615, "num_examples": 2961, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30504363, "num_examples": 5988, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15867991, "num_examples": 2961, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30600171, "num_examples": 5988, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 201153496, "size_in_bytes": 2576644039}, "cs_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 189543611, "num_examples": 58277, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27423365, "num_examples": 8284, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53790610, "num_examples": 16521, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16381360, "num_examples": 4920, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22244077, "num_examples": 6790, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 309383023, "size_in_bytes": 2684873566}, "cs_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 105431496, "num_examples": 35125, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15393683, "num_examples": 4943, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29975522, "num_examples": 9976, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15057384, "num_examples": 4820, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20541147, "num_examples": 6728, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 186399232, "size_in_bytes": 2561889775}, "cs_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 27816140, "num_examples": 9553, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4003326, "num_examples": 1327, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 7577366, "num_examples": 2636, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4024558, "num_examples": 1327, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7619542, "num_examples": 2636, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 51040932, "size_in_bytes": 2426531475}, "cs_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 90373560, "num_examples": 25474, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13325342, "num_examples": 3587, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 25528026, "num_examples": 7147, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13382734, "num_examples": 3587, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22210520, "num_examples": 6230, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 164820182, "size_in_bytes": 2540310725}, "cs_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 143538092, "num_examples": 48524, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20703526, "num_examples": 6887, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40469465, "num_examples": 13793, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14662779, "num_examples": 4794, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19894805, "num_examples": 6604, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 239268667, "size_in_bytes": 2614759210}, "cs_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 228385006, "num_examples": 49939, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33370534, "num_examples": 7148, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64614067, "num_examples": 14240, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22215975, "num_examples": 4860, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28209880, "num_examples": 6660, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 376795462, "size_in_bytes": 2752286005}, "cs_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 167808445, "num_examples": 54036, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24315076, "num_examples": 7673, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47560544, "num_examples": 15345, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15654992, "num_examples": 4840, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21072823, "num_examples": 6676, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 276411880, "size_in_bytes": 2651902423}, "cs_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 143186818, "num_examples": 46596, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20763438, "num_examples": 6595, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40499074, "num_examples": 13251, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15451066, "num_examples": 4848, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20866145, "num_examples": 6754, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 240766541, "size_in_bytes": 2616257084}, "cs_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 133555405, "num_examples": 33580, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 19595585, "num_examples": 4761, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 38126567, "num_examples": 9517, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19671761, "num_examples": 4761, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26141703, "num_examples": 6676, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 237091021, "size_in_bytes": 2612581564}, "cs_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 194291996, "num_examples": 70642, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28028213, "num_examples": 10065, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55164572, "num_examples": 20159, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14219366, "num_examples": 4900, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19649329, "num_examples": 6828, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 311353476, "size_in_bytes": 2686844019}, "cs_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 45397096, "num_examples": 15099, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6794491, "num_examples": 2154, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13085046, "num_examples": 4314, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6828955, "num_examples": 2154, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13154070, "num_examples": 4314, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 85259658, "size_in_bytes": 2460750201}, "cs_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 280766481, "num_examples": 93787, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40430724, "num_examples": 13388, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80243228, "num_examples": 26766, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15808560, "num_examples": 5072, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21994928, "num_examples": 7134, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 439243921, "size_in_bytes": 2814734464}, "hi_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 83267533, "num_examples": 16019, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11826763, "num_examples": 2201, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23175474, "num_examples": 4468, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11861979, "num_examples": 2201, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23246962, "num_examples": 4468, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 153378711, "size_in_bytes": 2528869254}, "hi_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 120763479, "num_examples": 24361, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17079340, "num_examples": 3437, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33245057, "num_examples": 6826, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17134332, "num_examples": 3437, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33352723, "num_examples": 6826, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 221574931, "size_in_bytes": 2597065474}, "hi_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 112212974, "num_examples": 27737, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16094331, "num_examples": 3883, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30899586, "num_examples": 7779, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16156459, "num_examples": 3883, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29178186, "num_examples": 7004, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 204541536, "size_in_bytes": 2580032079}, "hi_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 115199879, "num_examples": 23280, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15895116, "num_examples": 3195, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31560802, "num_examples": 6472, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15946236, "num_examples": 3195, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31664354, "num_examples": 6472, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 210266387, "size_in_bytes": 2585756930}, "hi_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169372685, "num_examples": 26493, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24408596, "num_examples": 3676, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46907083, "num_examples": 7442, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24467412, "num_examples": 3676, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47026155, "num_examples": 7442, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 312181931, "size_in_bytes": 2687672474}, "hi_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 250826482, "num_examples": 62878, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36130671, "num_examples": 8951, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 70848340, "num_examples": 17877, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25099338, "num_examples": 5587, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39323503, "num_examples": 8146, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 422228334, "size_in_bytes": 2797718877}, "hi_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 156765626, "num_examples": 38056, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22670001, "num_examples": 5362, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 44197964, "num_examples": 10820, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22340406, "num_examples": 5239, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34777093, "num_examples": 7572, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 280751090, "size_in_bytes": 2656241633}, "hi_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 75419659, "num_examples": 14316, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10649606, "num_examples": 2006, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20920574, "num_examples": 4074, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10681702, "num_examples": 2006, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20985758, "num_examples": 4074, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 138657299, "size_in_bytes": 2514147842}, "hi_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 151672398, "num_examples": 30717, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22250372, "num_examples": 4352, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 42533956, "num_examples": 8719, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22320004, "num_examples": 4352, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39241602, "num_examples": 7802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 278018332, "size_in_bytes": 2653508875}, "hi_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 198410095, "num_examples": 52143, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28621203, "num_examples": 7432, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55261956, "num_examples": 14757, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22589176, "num_examples": 5339, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34702720, "num_examples": 7568, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 339585150, "size_in_bytes": 2715075693}, "hi_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 286475401, "num_examples": 53992, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41468059, "num_examples": 7699, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80046422, "num_examples": 15318, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 30322316, "num_examples": 5411, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43659483, "num_examples": 7738, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 481971681, "size_in_bytes": 2857462224}, "hi_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 226297942, "num_examples": 58409, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32757408, "num_examples": 8314, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63773475, "num_examples": 16629, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24107580, "num_examples": 5481, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37306298, "num_examples": 7960, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 384242703, "size_in_bytes": 2759733246}, "hi_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 199816427, "num_examples": 50593, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28646700, "num_examples": 7134, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55736518, "num_examples": 14367, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23342952, "num_examples": 5387, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36121445, "num_examples": 7870, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 343664042, "size_in_bytes": 2719154585}, "hi_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 191212853, "num_examples": 37609, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27773477, "num_examples": 5302, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54062298, "num_examples": 10671, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27858309, "num_examples": 5302, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42095898, "num_examples": 7830, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 343002835, "size_in_bytes": 2718493378}, "hi_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 255707310, "num_examples": 75571, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36885455, "num_examples": 10790, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 72039266, "num_examples": 21519, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23088208, "num_examples": 5625, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36545783, "num_examples": 8188, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 424266022, "size_in_bytes": 2799756565}, "hi_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 112897266, "num_examples": 20826, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16397760, "num_examples": 2973, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31931211, "num_examples": 5952, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16445328, "num_examples": 2973, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32026443, "num_examples": 5952, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 209698008, "size_in_bytes": 2585188551}, "hi_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 344440176, "num_examples": 98666, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 49718675, "num_examples": 14125, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 97646232, "num_examples": 28108, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25108303, "num_examples": 5809, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39419404, "num_examples": 8476, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 556332790, "size_in_bytes": 2931823333}, "ja_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 58487808, "num_examples": 17942, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8444442, "num_examples": 2518, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16171454, "num_examples": 4942, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8484730, "num_examples": 2518, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16248976, "num_examples": 4942, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 107837410, "size_in_bytes": 2483327953}, "ja_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 120763479, "num_examples": 24361, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17079340, "num_examples": 3437, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33245057, "num_examples": 6826, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17134332, "num_examples": 3437, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33352723, "num_examples": 6826, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 221574931, "size_in_bytes": 2597065474}, "ja_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 87756583, "num_examples": 31780, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12344866, "num_examples": 4378, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24115343, "num_examples": 8843, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12414914, "num_examples": 4378, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22409417, "num_examples": 8068, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 159041123, "size_in_bytes": 2534531666}, "ja_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 88829224, "num_examples": 26663, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12310062, "num_examples": 3706, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24308366, "num_examples": 7394, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12369358, "num_examples": 3706, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24425120, "num_examples": 7394, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 162242130, "size_in_bytes": 2537732673}, "ja_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 143738159, "num_examples": 29880, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20320567, "num_examples": 4117, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39019776, "num_examples": 8214, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20386439, "num_examples": 4117, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39149650, "num_examples": 8214, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 262614591, "size_in_bytes": 2638105134}, "ja_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 224864836, "num_examples": 68323, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32223521, "num_examples": 9732, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63417618, "num_examples": 19413, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21204684, "num_examples": 6368, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31915807, "num_examples": 9682, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 373626466, "size_in_bytes": 2749117009}, "ja_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 132852290, "num_examples": 43007, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18910292, "num_examples": 5999, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36953492, "num_examples": 12104, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18590889, "num_examples": 5876, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27551615, "num_examples": 8856, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 234858578, "size_in_bytes": 2610349121}, "ja_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 51428301, "num_examples": 16339, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7246826, "num_examples": 2321, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 14100672, "num_examples": 4624, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7283962, "num_examples": 2321, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14173106, "num_examples": 4624, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 94232867, "size_in_bytes": 2469723410}, "ja_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 125152884, "num_examples": 35132, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17866457, "num_examples": 4901, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 34560178, "num_examples": 9797, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17944873, "num_examples": 4901, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31283522, "num_examples": 8880, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 226807914, "size_in_bytes": 2602298457}, "ja_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 172162050, "num_examples": 56686, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24456230, "num_examples": 8017, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47926259, "num_examples": 16047, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18433563, "num_examples": 5924, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27386113, "num_examples": 8858, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 290364215, "size_in_bytes": 2665854758}, "ja_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 262813872, "num_examples": 58599, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37747686, "num_examples": 8326, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73191221, "num_examples": 16596, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26611975, "num_examples": 6038, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36823180, "num_examples": 9016, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 437187934, "size_in_bytes": 2812678477}, "ja_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 200470407, "num_examples": 63584, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28744189, "num_examples": 9019, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 56516611, "num_examples": 18109, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20105641, "num_examples": 6186, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30071564, "num_examples": 9440, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 335908412, "size_in_bytes": 2711398955}, "ja_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 173342720, "num_examples": 55350, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24567346, "num_examples": 7761, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48332161, "num_examples": 15673, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19273630, "num_examples": 6014, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28736434, "num_examples": 9176, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 294252291, "size_in_bytes": 2669742834}, "ja_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166948680, "num_examples": 42488, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23640937, "num_examples": 5913, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46790553, "num_examples": 11919, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23735545, "num_examples": 5913, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34842571, "num_examples": 9078, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 295958286, "size_in_bytes": 2671448829}, "ja_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 228533279, "num_examples": 80952, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32636435, "num_examples": 11507, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64262307, "num_examples": 23013, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18850660, "num_examples": 6342, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28791178, "num_examples": 9682, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 373073859, "size_in_bytes": 2748564402}, "ja_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 82825334, "num_examples": 26559, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11845377, "num_examples": 3792, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23220805, "num_examples": 7590, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11906049, "num_examples": 3792, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23340695, "num_examples": 7590, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 153138260, "size_in_bytes": 2528628803}, "ja_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 316436712, "num_examples": 104203, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45292984, "num_examples": 14898, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89785881, "num_examples": 29710, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20694980, "num_examples": 6582, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31583135, "num_examples": 10078, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 503793692, "size_in_bytes": 2879284235}, "zh_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 61973073, "num_examples": 24110, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8973177, "num_examples": 3376, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 17408183, "num_examples": 6749, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9027193, "num_examples": 3376, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15670303, "num_examples": 5974, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 113051929, "size_in_bytes": 2488542472}, "zh_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 112212974, "num_examples": 27737, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16094331, "num_examples": 3883, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30899586, "num_examples": 7779, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16156459, "num_examples": 3883, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29178186, "num_examples": 7004, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 204541536, "size_in_bytes": 2580032079}, "zh_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 87756583, "num_examples": 31780, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12344866, "num_examples": 4378, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24115343, "num_examples": 8843, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12414914, "num_examples": 4378, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22409417, "num_examples": 8068, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 159041123, "size_in_bytes": 2534531666}, "zh_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 83386188, "num_examples": 30517, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11819410, "num_examples": 4226, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22859357, "num_examples": 8421, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11887026, "num_examples": 4226, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21148229, "num_examples": 7646, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 151100210, "size_in_bytes": 2526590753}, "zh_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 144866193, "num_examples": 35796, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20888047, "num_examples": 4925, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39733073, "num_examples": 9885, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20966847, "num_examples": 4925, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38045369, "num_examples": 9110, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 264499529, "size_in_bytes": 2639990072}, "zh_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 237933832, "num_examples": 79869, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34407713, "num_examples": 11348, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67170394, "num_examples": 22572, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23414732, "num_examples": 7984, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33701375, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 396628046, "size_in_bytes": 2772118589}, "zh_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 144755694, "num_examples": 53759, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20907033, "num_examples": 7521, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40506193, "num_examples": 15105, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20611982, "num_examples": 7398, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29308018, "num_examples": 11082, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 256088920, "size_in_bytes": 2631579463}, "zh_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 47397766, "num_examples": 19725, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6793425, "num_examples": 2779, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13146623, "num_examples": 5577, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6837889, "num_examples": 2779, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 11389991, "num_examples": 4802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 85565694, "size_in_bytes": 2461056237}, "zh_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 124818355, "num_examples": 41300, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18067256, "num_examples": 5723, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 34394974, "num_examples": 11372, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18158824, "num_examples": 5723, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29299204, "num_examples": 9680, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 224738613, "size_in_bytes": 2600229156}, "zh_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 184745590, "num_examples": 67666, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26437005, "num_examples": 9557, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 51484782, "num_examples": 19058, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20438978, "num_examples": 7464, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29148498, "num_examples": 11094, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 312254853, "size_in_bytes": 2687745396}, "zh_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 280825260, "num_examples": 69749, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40407301, "num_examples": 9856, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78794466, "num_examples": 19749, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29296070, "num_examples": 7568, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40632559, "num_examples": 11394, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 469955656, "size_in_bytes": 2845446199}, "zh_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 213998674, "num_examples": 74982, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 30773938, "num_examples": 10583, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 60510576, "num_examples": 21268, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22160414, "num_examples": 7750, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32271759, "num_examples": 11824, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 359715361, "size_in_bytes": 2735205904}, "zh_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 186466040, "num_examples": 66774, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26906549, "num_examples": 9421, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52400525, "num_examples": 18840, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21639393, "num_examples": 7674, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31011156, "num_examples": 11568, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 318423663, "size_in_bytes": 2693914206}, "zh_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 176483837, "num_examples": 51362, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25539348, "num_examples": 7225, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49710549, "num_examples": 14436, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25654948, "num_examples": 7225, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35958525, "num_examples": 10820, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 313347207, "size_in_bytes": 2688837750}, "zh_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 239533033, "num_examples": 91918, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34317307, "num_examples": 12997, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67387617, "num_examples": 26008, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20555372, "num_examples": 7832, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30120094, "num_examples": 11902, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 391913423, "size_in_bytes": 2767403966}, "zh_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 96675723, "num_examples": 39633, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 14153668, "num_examples": 5658, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 27544132, "num_examples": 11325, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14244196, "num_examples": 5658, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21896965, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 174514684, "size_in_bytes": 2550005227}, "zh_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 331493992, "num_examples": 116847, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47691542, "num_examples": 16692, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 94514334, "num_examples": 33345, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23122242, "num_examples": 8376, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32092411, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 528914521, "size_in_bytes": 2904405064}, "ko_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 59976457, "num_examples": 18463, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8491961, "num_examples": 2566, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16518964, "num_examples": 5074, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8533017, "num_examples": 2566, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16600148, "num_examples": 5074, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 110120547, "size_in_bytes": 2485611090}, "ko_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 115199879, "num_examples": 23280, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15895116, "num_examples": 3195, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31560802, "num_examples": 6472, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15946236, "num_examples": 3195, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31664354, "num_examples": 6472, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 210266387, "size_in_bytes": 2585756930}, "ko_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 88829224, "num_examples": 26663, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12310062, "num_examples": 3706, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24308366, "num_examples": 7394, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12369358, "num_examples": 3706, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24425120, "num_examples": 7394, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 162242130, "size_in_bytes": 2537732673}, "ko_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 83386188, "num_examples": 30517, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11819410, "num_examples": 4226, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22859357, "num_examples": 8421, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11887026, "num_examples": 4226, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21148229, "num_examples": 7646, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 151100210, "size_in_bytes": 2526590753}, "ko_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 148176444, "num_examples": 30373, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21018214, "num_examples": 4191, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40553060, "num_examples": 8432, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21085270, "num_examples": 4191, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40687972, "num_examples": 8432, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 271520960, "size_in_bytes": 2647011503}, "ko_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 224755039, "num_examples": 67910, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32474469, "num_examples": 9692, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63685470, "num_examples": 19315, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21454992, "num_examples": 6328, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32183641, "num_examples": 9584, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 374553611, "size_in_bytes": 2750044154}, "ko_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 131341149, "num_examples": 42294, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18700824, "num_examples": 5897, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36896546, "num_examples": 11932, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18379789, "num_examples": 5774, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27493467, "num_examples": 8684, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 232811775, "size_in_bytes": 2608302318}, "ko_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 49739881, "num_examples": 16184, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7090153, "num_examples": 2297, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13928138, "num_examples": 4626, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7126905, "num_examples": 2297, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14002154, "num_examples": 4626, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 91887231, "size_in_bytes": 2467377774}, "ko_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 129622264, "num_examples": 35965, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18695200, "num_examples": 5055, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36541317, "num_examples": 10177, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18776080, "num_examples": 5055, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33272291, "num_examples": 9260, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 236907152, "size_in_bytes": 2612397695}, "ko_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 170850671, "num_examples": 56081, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24327403, "num_examples": 7923, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48039863, "num_examples": 15927, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18303232, "num_examples": 5830, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27499347, "num_examples": 8738, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 289020516, "size_in_bytes": 2664511059}, "ko_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 260448901, "num_examples": 57712, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37504552, "num_examples": 8204, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73318249, "num_examples": 16430, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26366889, "num_examples": 5916, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36949102, "num_examples": 8850, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 434587693, "size_in_bytes": 2810078236}, "ko_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 197793930, "num_examples": 62449, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28283941, "num_examples": 8839, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55832536, "num_examples": 17761, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19642513, "num_examples": 6006, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29383471, "num_examples": 9092, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 330936391, "size_in_bytes": 2706426934}, "ko_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 173683965, "num_examples": 55197, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25097607, "num_examples": 7819, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48865451, "num_examples": 15705, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19804819, "num_examples": 6072, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29271786, "num_examples": 9208, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 296723628, "size_in_bytes": 2672214171}, "ko_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169182989, "num_examples": 42469, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24388096, "num_examples": 5993, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47723955, "num_examples": 11945, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24483984, "num_examples": 5993, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35777939, "num_examples": 9104, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 301556963, "size_in_bytes": 2677047506}, "ko_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 225683803, "num_examples": 79791, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32355907, "num_examples": 11353, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63767142, "num_examples": 22727, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18567668, "num_examples": 6188, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28292987, "num_examples": 9396, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 368667507, "size_in_bytes": 2744158050}, "ko_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81014576, "num_examples": 25572, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11682885, "num_examples": 3648, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22944915, "num_examples": 7308, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11741253, "num_examples": 3648, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23061843, "num_examples": 7308, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 150445472, "size_in_bytes": 2525936015}, "ko_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 315248680, "num_examples": 103456, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45358161, "num_examples": 14808, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89824786, "num_examples": 29528, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20758717, "num_examples": 6492, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31620678, "num_examples": 9896, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 502811022, "size_in_bytes": 2878301565}, "th_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 108360356, "num_examples": 21378, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15820615, "num_examples": 2961, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30504363, "num_examples": 5988, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15867991, "num_examples": 2961, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30600171, "num_examples": 5988, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 201153496, "size_in_bytes": 2576644039}, "th_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169372685, "num_examples": 26493, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24408596, "num_examples": 3676, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46907083, "num_examples": 7442, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24467412, "num_examples": 3676, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47026155, "num_examples": 7442, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 312181931, "size_in_bytes": 2687672474}, "th_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 143738159, "num_examples": 29880, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20320567, "num_examples": 4117, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39019776, "num_examples": 8214, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20386439, "num_examples": 4117, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39149650, "num_examples": 8214, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 262614591, "size_in_bytes": 2638105134}, "th_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 144866193, "num_examples": 35796, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20888047, "num_examples": 4925, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39733073, "num_examples": 9885, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20966847, "num_examples": 4925, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38045369, "num_examples": 9110, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 264499529, "size_in_bytes": 2639990072}, "th_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 148176444, "num_examples": 30373, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21018214, "num_examples": 4191, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40553060, "num_examples": 8432, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21085270, "num_examples": 4191, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40687972, "num_examples": 8432, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 271520960, "size_in_bytes": 2647011503}, "th_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 293898448, "num_examples": 72439, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42796665, "num_examples": 10287, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 82741769, "num_examples": 20509, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31786708, "num_examples": 6923, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51259044, "num_examples": 10778, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 502482634, "size_in_bytes": 2877973177}, "th_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 199289191, "num_examples": 47499, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28833817, "num_examples": 6594, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55376986, "num_examples": 13306, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28523934, "num_examples": 6471, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45995891, "num_examples": 10058, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 358019819, "size_in_bytes": 2733510362}, "th_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 89830305, "num_examples": 17859, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13127248, "num_examples": 2526, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 25179461, "num_examples": 5116, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13167664, "num_examples": 2526, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25261317, "num_examples": 5116, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 166565995, "size_in_bytes": 2542056538}, "th_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 198724004, "num_examples": 40888, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 29119772, "num_examples": 5726, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55468764, "num_examples": 11487, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29211388, "num_examples": 5726, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 52220698, "num_examples": 10570, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 364744626, "size_in_bytes": 2740235169}, "th_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 234778548, "num_examples": 60546, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33895191, "num_examples": 8524, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65422352, "num_examples": 17075, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27880636, "num_examples": 6431, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44900204, "num_examples": 9886, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 406876931, "size_in_bytes": 2782367474}, "th_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 327187391, "num_examples": 62373, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47422031, "num_examples": 8797, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91632294, "num_examples": 17664, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36293856, "num_examples": 6509, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 55282891, "num_examples": 10084, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 557818463, "size_in_bytes": 2933309006}, "th_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 265992739, "num_examples": 67492, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 38571672, "num_examples": 9520, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 74828982, "num_examples": 19137, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29941140, "num_examples": 6687, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 48401933, "num_examples": 10468, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 457736466, "size_in_bytes": 2833227009}, "th_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 241322806, "num_examples": 60048, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34933713, "num_examples": 8434, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67211234, "num_examples": 16975, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29650765, "num_examples": 6687, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47637889, "num_examples": 10478, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 420756407, "size_in_bytes": 2796246950}, "th_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 241944775, "num_examples": 47948, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35190181, "num_examples": 6698, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 68345539, "num_examples": 13537, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35297349, "num_examples": 6698, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 56424995, "num_examples": 10696, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 437202839, "size_in_bytes": 2812693382}, "th_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 294294528, "num_examples": 84730, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42635895, "num_examples": 11994, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 82965864, "num_examples": 24067, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28857912, "num_examples": 6829, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47513149, "num_examples": 10736, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 496267348, "size_in_bytes": 2871757891}, "th_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 155128871, "num_examples": 30975, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23124256, "num_examples": 4425, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 43977504, "num_examples": 8850, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23195056, "num_examples": 4425, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44119104, "num_examples": 8850, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 289544791, "size_in_bytes": 2665035334}, "th_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 387507666, "num_examples": 108637, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56359166, "num_examples": 15523, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 109985100, "num_examples": 30940, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31771162, "num_examples": 7207, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51803584, "num_examples": 11308, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 637426678, "size_in_bytes": 3012917221}, "fr_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 189543611, "num_examples": 58277, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27423365, "num_examples": 8284, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53790610, "num_examples": 16521, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16381360, "num_examples": 4920, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22244077, "num_examples": 6790, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 309383023, "size_in_bytes": 2684873566}, "fr_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 250826482, "num_examples": 62878, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36130671, "num_examples": 8951, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 70848340, "num_examples": 17877, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25099338, "num_examples": 5587, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39323503, "num_examples": 8146, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 422228334, "size_in_bytes": 2797718877}, "fr_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 224864836, "num_examples": 68323, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32223521, "num_examples": 9732, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63417618, "num_examples": 19413, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21204684, "num_examples": 6368, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31915807, "num_examples": 9682, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 373626466, "size_in_bytes": 2749117009}, "fr_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 237933832, "num_examples": 79869, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34407713, "num_examples": 11348, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67170394, "num_examples": 22572, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23414732, "num_examples": 7984, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33701375, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 396628046, "size_in_bytes": 2772118589}, "fr_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 224755039, "num_examples": 67910, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32474469, "num_examples": 9692, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63685470, "num_examples": 19315, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21454992, "num_examples": 6328, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32183641, "num_examples": 9584, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 374553611, "size_in_bytes": 2750044154}, "fr_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 293898448, "num_examples": 72439, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42796665, "num_examples": 10287, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 82741769, "num_examples": 20509, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31786708, "num_examples": 6923, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51259044, "num_examples": 10778, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 502482634, "size_in_bytes": 2877973177}, "fr_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 328025549, "num_examples": 104090, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47237053, "num_examples": 14663, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91757469, "num_examples": 29179, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35881725, "num_examples": 11176, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37585145, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 540486941, "size_in_bytes": 2915977484}, "fr_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169657827, "num_examples": 52964, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24489928, "num_examples": 7577, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48191376, "num_examples": 15131, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13436611, "num_examples": 4213, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16622603, "num_examples": 5400, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 272398345, "size_in_bytes": 2647888888}, "fr_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 283378423, "num_examples": 81891, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41169674, "num_examples": 11613, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 79915832, "num_examples": 23100, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 30180933, "num_examples": 8249, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43220661, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 477865523, "size_in_bytes": 2853356066}, "fr_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 417881614, "num_examples": 134021, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56582063, "num_examples": 17877, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 111206996, "num_examples": 35742, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38134748, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37621447, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 661426868, "size_in_bytes": 3036917411}, "fr_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 556537507, "num_examples": 137322, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 74694956, "num_examples": 18140, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 145307684, "num_examples": 36035, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 50492226, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49551148, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 876583521, "size_in_bytes": 3252074064}, "fr_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 488017961, "num_examples": 152991, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 65921178, "num_examples": 20327, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129677763, "num_examples": 40510, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39157742, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38747621, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 761522265, "size_in_bytes": 3137012808}, "fr_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 427216106, "num_examples": 133653, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 60456654, "num_examples": 18625, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 118392495, "num_examples": 37170, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39189128, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38380590, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 683634973, "size_in_bytes": 3059125516}, "fr_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 368679326, "num_examples": 100111, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 53272378, "num_examples": 14189, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 104133848, "num_examples": 28172, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42324853, "num_examples": 10825, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46594170, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 615004575, "size_in_bytes": 2990495118}, "fr_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 515882894, "num_examples": 173551, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 68794446, "num_examples": 22831, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 134550579, "num_examples": 45266, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36848981, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36126524, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 792203424, "size_in_bytes": 3167693967}, "fr_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 411286458, "num_examples": 133668, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59349317, "num_examples": 19092, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 117512924, "num_examples": 38193, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28068752, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28260502, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 644477953, "size_in_bytes": 3019968496}, "fr_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 644793465, "num_examples": 207080, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 91017234, "num_examples": 28904, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 179593516, "num_examples": 57533, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38444271, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38352312, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 992200798, "size_in_bytes": 3367691341}, "nl_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 105431496, "num_examples": 35125, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15393683, "num_examples": 4943, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29975522, "num_examples": 9976, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15057384, "num_examples": 4820, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20541147, "num_examples": 6728, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 186399232, "size_in_bytes": 2561889775}, "nl_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 156765626, "num_examples": 38056, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22670001, "num_examples": 5362, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 44197964, "num_examples": 10820, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22340406, "num_examples": 5239, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34777093, "num_examples": 7572, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 280751090, "size_in_bytes": 2656241633}, "nl_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 132852290, "num_examples": 43007, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18910292, "num_examples": 5999, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36953492, "num_examples": 12104, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18590889, "num_examples": 5876, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27551615, "num_examples": 8856, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 234858578, "size_in_bytes": 2610349121}, "nl_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 144755694, "num_examples": 53759, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20907033, "num_examples": 7521, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40506193, "num_examples": 15105, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20611982, "num_examples": 7398, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29308018, "num_examples": 11082, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 256088920, "size_in_bytes": 2631579463}, "nl_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 131341149, "num_examples": 42294, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18700824, "num_examples": 5897, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36896546, "num_examples": 11932, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18379789, "num_examples": 5774, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27493467, "num_examples": 8684, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 232811775, "size_in_bytes": 2608302318}, "nl_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 199289191, "num_examples": 47499, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28833817, "num_examples": 6594, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55376986, "num_examples": 13306, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28523934, "num_examples": 6471, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45995891, "num_examples": 10058, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 358019819, "size_in_bytes": 2733510362}, "nl_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 328025549, "num_examples": 104090, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47237053, "num_examples": 14663, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91757469, "num_examples": 29179, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35881725, "num_examples": 11176, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37585145, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 540486941, "size_in_bytes": 2915977484}, "nl_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 83105825, "num_examples": 28648, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11967030, "num_examples": 4042, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23375639, "num_examples": 8152, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11616315, "num_examples": 3919, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13912080, "num_examples": 4904, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 143976889, "size_in_bytes": 2519467432}, "nl_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 181636587, "num_examples": 54869, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26462728, "num_examples": 7668, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 50810986, "num_examples": 15371, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26170029, "num_examples": 7545, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38031073, "num_examples": 11206, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 323111403, "size_in_bytes": 2698601946}, "nl_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 253874654, "num_examples": 86377, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35748097, "num_examples": 11912, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69699454, "num_examples": 23897, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29372363, "num_examples": 9696, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35333724, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 424028292, "size_in_bytes": 2799518835}, "nl_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 363222499, "num_examples": 88734, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 51507709, "num_examples": 12271, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 99795847, "num_examples": 24612, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 40019731, "num_examples": 9860, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46994280, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 601540066, "size_in_bytes": 2977030609}, "nl_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 297517128, "num_examples": 97819, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42449172, "num_examples": 13644, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 82937447, "num_examples": 27377, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33469237, "num_examples": 10688, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36446196, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 492819180, "size_in_bytes": 2868309723}, "nl_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 268310694, "num_examples": 89055, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 38170018, "num_examples": 12354, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 74184620, "num_examples": 24805, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 32534403, "num_examples": 10484, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35891073, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 449090808, "size_in_bytes": 2824581351}, "nl_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 263048332, "num_examples": 73181, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37565756, "num_examples": 10176, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73717887, "num_examples": 20483, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37313185, "num_examples": 10053, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43779436, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 455424596, "size_in_bytes": 2830915139}, "nl_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 320040173, "num_examples": 114523, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45452674, "num_examples": 15998, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89143645, "num_examples": 32097, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31323368, "num_examples": 10710, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33925525, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 519885385, "size_in_bytes": 2895375928}, "nl_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 188073112, "num_examples": 65598, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27612857, "num_examples": 9369, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53427535, "num_examples": 18744, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26562758, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25844976, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 321521238, "size_in_bytes": 2697011781}, "nl_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 422075349, "num_examples": 141670, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 60963067, "num_examples": 20183, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 119689771, "num_examples": 40212, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36034236, "num_examples": 11744, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35983844, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 674746267, "size_in_bytes": 3050236810}, "tr_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 27816140, "num_examples": 9553, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4003326, "num_examples": 1327, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 7577366, "num_examples": 2636, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4024558, "num_examples": 1327, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7619542, "num_examples": 2636, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 51040932, "size_in_bytes": 2426531475}, "tr_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 75419659, "num_examples": 14316, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10649606, "num_examples": 2006, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20920574, "num_examples": 4074, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10681702, "num_examples": 2006, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20985758, "num_examples": 4074, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 138657299, "size_in_bytes": 2514147842}, "tr_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 51428301, "num_examples": 16339, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7246826, "num_examples": 2321, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 14100672, "num_examples": 4624, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7283962, "num_examples": 2321, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14173106, "num_examples": 4624, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 94232867, "size_in_bytes": 2469723410}, "tr_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 47397766, "num_examples": 19725, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6793425, "num_examples": 2779, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13146623, "num_examples": 5577, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6837889, "num_examples": 2779, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 11389991, "num_examples": 4802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 85565694, "size_in_bytes": 2461056237}, "tr_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 49739881, "num_examples": 16184, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7090153, "num_examples": 2297, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13928138, "num_examples": 4626, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7126905, "num_examples": 2297, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14002154, "num_examples": 4626, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 91887231, "size_in_bytes": 2467377774}, "tr_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 89830305, "num_examples": 17859, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13127248, "num_examples": 2526, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 25179461, "num_examples": 5116, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13167664, "num_examples": 2526, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25261317, "num_examples": 5116, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 166565995, "size_in_bytes": 2542056538}, "tr_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169657827, "num_examples": 52964, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24489928, "num_examples": 7577, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48191376, "num_examples": 15131, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13436611, "num_examples": 4213, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16622603, "num_examples": 5400, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 272398345, "size_in_bytes": 2647888888}, "tr_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 83105825, "num_examples": 28648, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11967030, "num_examples": 4042, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23375639, "num_examples": 8152, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11616315, "num_examples": 3919, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13912080, "num_examples": 4904, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 143976889, "size_in_bytes": 2519467432}, "tr_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 74215809, "num_examples": 21537, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10944113, "num_examples": 3082, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20926222, "num_examples": 6129, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10993425, "num_examples": 3082, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 17592428, "num_examples": 5212, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 134671997, "size_in_bytes": 2510162540}, "tr_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 125876720, "num_examples": 43499, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18054248, "num_examples": 6226, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 35227253, "num_examples": 12397, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12002925, "num_examples": 4133, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14630257, "num_examples": 5208, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 205791403, "size_in_bytes": 2581281946}, "tr_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 206832302, "num_examples": 45064, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 29791308, "num_examples": 6421, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 58084198, "num_examples": 12874, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18625117, "num_examples": 4133, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21658155, "num_examples": 5294, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 334991080, "size_in_bytes": 2710481623}, "tr_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 148317004, "num_examples": 48849, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21270709, "num_examples": 6942, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 41901565, "num_examples": 13955, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12598929, "num_examples": 4109, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15391604, "num_examples": 5286, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 239479811, "size_in_bytes": 2614970354}, "tr_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 123687320, "num_examples": 41313, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17827957, "num_examples": 5884, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 34799859, "num_examples": 11801, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12504209, "num_examples": 4137, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15143730, "num_examples": 5304, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 203963075, "size_in_bytes": 2579453618}, "tr_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 110602032, "num_examples": 27899, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15960618, "num_examples": 3944, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31408171, "num_examples": 7941, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16023722, "num_examples": 3944, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19398091, "num_examples": 5100, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 193392634, "size_in_bytes": 2568883177}, "tr_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 176432382, "num_examples": 65719, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25253786, "num_examples": 9376, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49706176, "num_examples": 18765, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11433915, "num_examples": 4211, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14168629, "num_examples": 5434, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 276994888, "size_in_bytes": 2652485431}, "tr_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 25008753, "num_examples": 9444, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 3560378, "num_examples": 1347, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 6853199, "num_examples": 2700, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 3581930, "num_examples": 1347, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 6896399, "num_examples": 2700, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 45900659, "size_in_bytes": 2421391202}, "tr_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 259722926, "num_examples": 87986, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37191372, "num_examples": 12591, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73811477, "num_examples": 25112, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12556456, "num_examples": 4275, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15536713, "num_examples": 5480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 398818944, "size_in_bytes": 2774309487}, "vi_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 90373560, "num_examples": 25474, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13325342, "num_examples": 3587, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 25528026, "num_examples": 7147, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13382734, "num_examples": 3587, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22210520, "num_examples": 6230, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 164820182, "size_in_bytes": 2540310725}, "vi_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 151672398, "num_examples": 30717, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22250372, "num_examples": 4352, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 42533956, "num_examples": 8719, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22320004, "num_examples": 4352, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39241602, "num_examples": 7802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 278018332, "size_in_bytes": 2653508875}, "vi_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 125152884, "num_examples": 35132, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17866457, "num_examples": 4901, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 34560178, "num_examples": 9797, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17944873, "num_examples": 4901, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31283522, "num_examples": 8880, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 226807914, "size_in_bytes": 2602298457}, "vi_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 124818355, "num_examples": 41300, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18067256, "num_examples": 5723, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 34394974, "num_examples": 11372, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18158824, "num_examples": 5723, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29299204, "num_examples": 9680, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 224738613, "size_in_bytes": 2600229156}, "vi_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 129622264, "num_examples": 35965, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18695200, "num_examples": 5055, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36541317, "num_examples": 10177, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18776080, "num_examples": 5055, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33272291, "num_examples": 9260, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 236907152, "size_in_bytes": 2612397695}, "vi_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 198724004, "num_examples": 40888, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 29119772, "num_examples": 5726, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55468764, "num_examples": 11487, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29211388, "num_examples": 5726, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 52220698, "num_examples": 10570, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 364744626, "size_in_bytes": 2740235169}, "vi_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 283378423, "num_examples": 81891, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41169674, "num_examples": 11613, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 79915832, "num_examples": 23100, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 30180933, "num_examples": 8249, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43220661, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 477865523, "size_in_bytes": 2853356066}, "vi_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 181636587, "num_examples": 54869, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26462728, "num_examples": 7668, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 50810986, "num_examples": 15371, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26170029, "num_examples": 7545, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38031073, "num_examples": 11206, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 323111403, "size_in_bytes": 2698601946}, "vi_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 74215809, "num_examples": 21537, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10944113, "num_examples": 3082, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20926222, "num_examples": 6129, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10993425, "num_examples": 3082, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 17592428, "num_examples": 5212, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 134671997, "size_in_bytes": 2510162540}, "vi_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 219166869, "num_examples": 67890, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31778960, "num_examples": 9594, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 61212341, "num_examples": 19086, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25781525, "num_examples": 7501, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37290511, "num_examples": 10980, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 375230206, "size_in_bytes": 2750720749}, "vi_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 317107402, "num_examples": 69761, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46016881, "num_examples": 9859, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89186447, "num_examples": 19741, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34905698, "num_examples": 7571, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49438418, "num_examples": 11244, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 536654846, "size_in_bytes": 2912145389}, "vi_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 252112751, "num_examples": 75600, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36443765, "num_examples": 10666, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 70958519, "num_examples": 21352, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27831569, "num_examples": 7833, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 41135052, "num_examples": 11766, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 428481656, "size_in_bytes": 2803972199}, "vi_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 227709647, "num_examples": 68428, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33083594, "num_examples": 9638, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63938419, "num_examples": 19302, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27819910, "num_examples": 7891, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40970448, "num_examples": 11888, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 393522018, "size_in_bytes": 2769012561}, "vi_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 231959596, "num_examples": 56694, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33705379, "num_examples": 7956, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65600585, "num_examples": 15962, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33832675, "num_examples": 7956, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49509101, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 414607336, "size_in_bytes": 2790097879}, "vi_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 279114345, "num_examples": 93098, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40260120, "num_examples": 13174, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78808326, "num_examples": 26370, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26501017, "num_examples": 8009, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39569743, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 464253551, "size_in_bytes": 2839744094}, "vi_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 142275998, "num_examples": 41121, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21099930, "num_examples": 5871, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40754049, "num_examples": 11751, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21193866, "num_examples": 5871, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31471022, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 256794865, "size_in_bytes": 2632285408}, "vi_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 375252849, "num_examples": 118093, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 54466998, "num_examples": 16897, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 106888168, "num_examples": 33617, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29900978, "num_examples": 8581, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 41566919, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 608075912, "size_in_bytes": 2983566455}, "it_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 143538092, "num_examples": 48524, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20703526, "num_examples": 6887, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40469465, "num_examples": 13793, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14662779, "num_examples": 4794, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19894805, "num_examples": 6604, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 239268667, "size_in_bytes": 2614759210}, "it_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 198410095, "num_examples": 52143, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28621203, "num_examples": 7432, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55261956, "num_examples": 14757, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22589176, "num_examples": 5339, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34702720, "num_examples": 7568, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 339585150, "size_in_bytes": 2715075693}, "it_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 172162050, "num_examples": 56686, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24456230, "num_examples": 8017, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47926259, "num_examples": 16047, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18433563, "num_examples": 5924, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27386113, "num_examples": 8858, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 290364215, "size_in_bytes": 2665854758}, "it_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 184745590, "num_examples": 67666, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26437005, "num_examples": 9557, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 51484782, "num_examples": 19058, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20438978, "num_examples": 7464, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29148498, "num_examples": 11094, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 312254853, "size_in_bytes": 2687745396}, "it_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 170850671, "num_examples": 56081, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24327403, "num_examples": 7923, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48039863, "num_examples": 15927, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18303232, "num_examples": 5830, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27499347, "num_examples": 8738, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 289020516, "size_in_bytes": 2664511059}, "it_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 234778548, "num_examples": 60546, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33895191, "num_examples": 8524, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65422352, "num_examples": 17075, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27880636, "num_examples": 6431, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44900204, "num_examples": 9886, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 406876931, "size_in_bytes": 2782367474}, "it_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 417881614, "num_examples": 134021, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56582063, "num_examples": 17877, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 111206996, "num_examples": 35742, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38134748, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37621447, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 661426868, "size_in_bytes": 3036917411}, "it_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 253874654, "num_examples": 86377, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35748097, "num_examples": 11912, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69699454, "num_examples": 23897, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29372363, "num_examples": 9696, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35333724, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 424028292, "size_in_bytes": 2799518835}, "it_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 125876720, "num_examples": 43499, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18054248, "num_examples": 6226, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 35227253, "num_examples": 12397, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12002925, "num_examples": 4133, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14630257, "num_examples": 5208, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 205791403, "size_in_bytes": 2581281946}, "it_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 219166869, "num_examples": 67890, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31778960, "num_examples": 9594, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 61212341, "num_examples": 19086, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25781525, "num_examples": 7501, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37290511, "num_examples": 10980, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 375230206, "size_in_bytes": 2750720749}, "it_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 480295607, "num_examples": 123167, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64720404, "num_examples": 16383, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 125355954, "num_examples": 32589, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 47556909, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46002385, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 763931259, "size_in_bytes": 3139421802}, "it_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 386400628, "num_examples": 128404, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 51920107, "num_examples": 17082, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 101970017, "num_examples": 34120, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36733200, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35974957, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 612998909, "size_in_bytes": 2988489452}, "it_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 335081308, "num_examples": 111560, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45965925, "num_examples": 15050, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89483518, "num_examples": 30068, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34637894, "num_examples": 11210, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36012632, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 541181277, "size_in_bytes": 2916671820}, "it_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 289560780, "num_examples": 82842, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41015302, "num_examples": 11496, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80623597, "num_examples": 23118, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35048299, "num_examples": 9403, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44008429, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 490256407, "size_in_bytes": 2865746950}, "it_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 426623553, "num_examples": 152968, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59201541, "num_examples": 21130, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 116270099, "num_examples": 42252, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34239466, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32997725, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 669332384, "size_in_bytes": 3044822927}, "it_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 300597006, "num_examples": 106983, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 43061023, "num_examples": 15279, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 84501587, "num_examples": 30567, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25513996, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24859349, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 478532961, "size_in_bytes": 2854023504}, "it_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 536239898, "num_examples": 181897, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 76139890, "num_examples": 25749, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 150448276, "num_examples": 51463, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35603354, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35018317, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 833449735, "size_in_bytes": 3208940278}, "ru_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 228385006, "num_examples": 49939, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33370534, "num_examples": 7148, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64614067, "num_examples": 14240, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22215975, "num_examples": 4860, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28209880, "num_examples": 6660, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 376795462, "size_in_bytes": 2752286005}, "ru_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 286475401, "num_examples": 53992, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41468059, "num_examples": 7699, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80046422, "num_examples": 15318, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 30322316, "num_examples": 5411, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43659483, "num_examples": 7738, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 481971681, "size_in_bytes": 2857462224}, "ru_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 262813872, "num_examples": 58599, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37747686, "num_examples": 8326, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73191221, "num_examples": 16596, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26611975, "num_examples": 6038, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36823180, "num_examples": 9016, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 437187934, "size_in_bytes": 2812678477}, "ru_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 280825260, "num_examples": 69749, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40407301, "num_examples": 9856, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78794466, "num_examples": 19749, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29296070, "num_examples": 7568, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40632559, "num_examples": 11394, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 469955656, "size_in_bytes": 2845446199}, "ru_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 260448901, "num_examples": 57712, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37504552, "num_examples": 8204, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73318249, "num_examples": 16430, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26366889, "num_examples": 5916, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36949102, "num_examples": 8850, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 434587693, "size_in_bytes": 2810078236}, "ru_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 327187391, "num_examples": 62373, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47422031, "num_examples": 8797, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91632294, "num_examples": 17664, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36293856, "num_examples": 6509, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 55282891, "num_examples": 10084, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 557818463, "size_in_bytes": 2933309006}, "ru_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 556537507, "num_examples": 137322, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 74694956, "num_examples": 18140, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 145307684, "num_examples": 36035, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 50492226, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49551148, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 876583521, "size_in_bytes": 3252074064}, "ru_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 363222499, "num_examples": 88734, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 51507709, "num_examples": 12271, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 99795847, "num_examples": 24612, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 40019731, "num_examples": 9860, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46994280, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 601540066, "size_in_bytes": 2977030609}, "ru_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 206832302, "num_examples": 45064, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 29791308, "num_examples": 6421, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 58084198, "num_examples": 12874, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18625117, "num_examples": 4133, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21658155, "num_examples": 5294, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 334991080, "size_in_bytes": 2710481623}, "ru_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 317107402, "num_examples": 69761, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46016881, "num_examples": 9859, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89186447, "num_examples": 19741, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34905698, "num_examples": 7571, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49438418, "num_examples": 11244, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 536654846, "size_in_bytes": 2912145389}, "ru_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 480295607, "num_examples": 123167, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64720404, "num_examples": 16383, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 125355954, "num_examples": 32589, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 47556909, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46002385, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 763931259, "size_in_bytes": 3139421802}, "ru_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 521353577, "num_examples": 131497, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 69373926, "num_examples": 17217, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 134623372, "num_examples": 34219, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 49203297, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47762670, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 822316842, "size_in_bytes": 3197807385}, "ru_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 460304514, "num_examples": 114385, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 63424862, "num_examples": 15495, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 122858084, "num_examples": 30913, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46985963, "num_examples": 11460, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47418902, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 740992325, "size_in_bytes": 3116482868}, "ru_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 396054199, "num_examples": 84835, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56567853, "num_examples": 11857, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 110756593, "num_examples": 23719, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 45488638, "num_examples": 9569, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 56334575, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 665201858, "size_in_bytes": 3040692401}, "ru_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 566144695, "num_examples": 154721, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 77493111, "num_examples": 20973, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 152245651, "num_examples": 42067, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46579271, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45276455, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 887739183, "size_in_bytes": 3263229726}, "ru_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 459684728, "num_examples": 111084, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 66397480, "num_examples": 15864, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129667377, "num_examples": 31740, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37928230, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36795657, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 730473472, "size_in_bytes": 3105964015}, "ru_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 685054819, "num_examples": 184296, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 97093934, "num_examples": 25888, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 190739233, "num_examples": 51730, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 48206822, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47058922, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 1068153730, "size_in_bytes": 3443644273}, "de_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 167808445, "num_examples": 54036, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24315076, "num_examples": 7673, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47560544, "num_examples": 15345, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15654992, "num_examples": 4840, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21072823, "num_examples": 6676, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 276411880, "size_in_bytes": 2651902423}, "de_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 226297942, "num_examples": 58409, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32757408, "num_examples": 8314, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63773475, "num_examples": 16629, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24107580, "num_examples": 5481, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37306298, "num_examples": 7960, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 384242703, "size_in_bytes": 2759733246}, "de_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 200470407, "num_examples": 63584, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28744189, "num_examples": 9019, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 56516611, "num_examples": 18109, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20105641, "num_examples": 6186, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30071564, "num_examples": 9440, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 335908412, "size_in_bytes": 2711398955}, "de_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 213998674, "num_examples": 74982, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 30773938, "num_examples": 10583, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 60510576, "num_examples": 21268, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22160414, "num_examples": 7750, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32271759, "num_examples": 11824, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 359715361, "size_in_bytes": 2735205904}, "de_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 197793930, "num_examples": 62449, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28283941, "num_examples": 8839, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55832536, "num_examples": 17761, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19642513, "num_examples": 6006, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29383471, "num_examples": 9092, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 330936391, "size_in_bytes": 2706426934}, "de_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 265992739, "num_examples": 67492, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 38571672, "num_examples": 9520, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 74828982, "num_examples": 19137, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29941140, "num_examples": 6687, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 48401933, "num_examples": 10468, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 457736466, "size_in_bytes": 2833227009}, "de_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 488017961, "num_examples": 152991, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 65921178, "num_examples": 20327, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129677763, "num_examples": 40510, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39157742, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38747621, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 761522265, "size_in_bytes": 3137012808}, "de_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 297517128, "num_examples": 97819, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42449172, "num_examples": 13644, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 82937447, "num_examples": 27377, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33469237, "num_examples": 10688, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36446196, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 492819180, "size_in_bytes": 2868309723}, "de_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 148317004, "num_examples": 48849, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21270709, "num_examples": 6942, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 41901565, "num_examples": 13955, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12598929, "num_examples": 4109, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15391604, "num_examples": 5286, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 239479811, "size_in_bytes": 2614970354}, "de_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 252112751, "num_examples": 75600, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36443765, "num_examples": 10666, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 70958519, "num_examples": 21352, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27831569, "num_examples": 7833, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 41135052, "num_examples": 11766, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 428481656, "size_in_bytes": 2803972199}, "de_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 386400628, "num_examples": 128404, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 51920107, "num_examples": 17082, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 101970017, "num_examples": 34120, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36733200, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35974957, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 612998909, "size_in_bytes": 2988489452}, "de_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 521353577, "num_examples": 131497, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 69373926, "num_examples": 17217, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 134623372, "num_examples": 34219, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 49203297, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47762670, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 822316842, "size_in_bytes": 3197807385}, "de_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 386150473, "num_examples": 124866, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 53825037, "num_examples": 17144, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 105282698, "num_examples": 34300, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37985941, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37017309, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 620261458, "size_in_bytes": 2995752001}, "de_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 332912886, "num_examples": 93020, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47629352, "num_examples": 13056, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 93632278, "num_examples": 26144, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39055396, "num_examples": 10223, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45226891, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 558456803, "size_in_bytes": 2933947346}, "de_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 474936699, "num_examples": 165092, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64418438, "num_examples": 22196, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 126646414, "num_examples": 44278, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35609586, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34748803, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 736359940, "size_in_bytes": 3111850483}, "de_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 360916674, "num_examples": 122517, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 52048916, "num_examples": 17499, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 102578311, "num_examples": 35007, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27172663, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26585059, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 569301623, "size_in_bytes": 2944792166}, "de_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 595260381, "num_examples": 196565, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 84007562, "num_examples": 27557, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 166264097, "num_examples": 54989, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37244632, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36783845, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 919560517, "size_in_bytes": 3295051060}, "id_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 143186818, "num_examples": 46596, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20763438, "num_examples": 6595, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40499074, "num_examples": 13251, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15451066, "num_examples": 4848, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20866145, "num_examples": 6754, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 240766541, "size_in_bytes": 2616257084}, "id_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 199816427, "num_examples": 50593, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28646700, "num_examples": 7134, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55736518, "num_examples": 14367, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23342952, "num_examples": 5387, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36121445, "num_examples": 7870, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 343664042, "size_in_bytes": 2719154585}, "id_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 173342720, "num_examples": 55350, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24567346, "num_examples": 7761, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48332161, "num_examples": 15673, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19273630, "num_examples": 6014, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28736434, "num_examples": 9176, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 294252291, "size_in_bytes": 2669742834}, "id_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 186466040, "num_examples": 66774, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26906549, "num_examples": 9421, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52400525, "num_examples": 18840, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21639393, "num_examples": 7674, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31011156, "num_examples": 11568, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 318423663, "size_in_bytes": 2693914206}, "id_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 173683965, "num_examples": 55197, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25097607, "num_examples": 7819, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48865451, "num_examples": 15705, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19804819, "num_examples": 6072, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29271786, "num_examples": 9208, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 296723628, "size_in_bytes": 2672214171}, "id_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 241322806, "num_examples": 60048, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34933713, "num_examples": 8434, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67211234, "num_examples": 16975, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29650765, "num_examples": 6687, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47637889, "num_examples": 10478, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 420756407, "size_in_bytes": 2796246950}, "id_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 427216106, "num_examples": 133653, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 60456654, "num_examples": 18625, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 118392495, "num_examples": 37170, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39189128, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38380590, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 683634973, "size_in_bytes": 3059125516}, "id_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 268310694, "num_examples": 89055, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 38170018, "num_examples": 12354, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 74184620, "num_examples": 24805, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 32534403, "num_examples": 10484, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35891073, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 449090808, "size_in_bytes": 2824581351}, "id_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 123687320, "num_examples": 41313, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17827957, "num_examples": 5884, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 34799859, "num_examples": 11801, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12504209, "num_examples": 4137, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15143730, "num_examples": 5304, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 203963075, "size_in_bytes": 2579453618}, "id_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 227709647, "num_examples": 68428, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33083594, "num_examples": 9638, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63938419, "num_examples": 19302, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27819910, "num_examples": 7891, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40970448, "num_examples": 11888, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 393522018, "size_in_bytes": 2769012561}, "id_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 335081308, "num_examples": 111560, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45965925, "num_examples": 15050, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89483518, "num_examples": 30068, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34637894, "num_examples": 11210, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36012632, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 541181277, "size_in_bytes": 2916671820}, "id_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 460304514, "num_examples": 114385, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 63424862, "num_examples": 15495, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 122858084, "num_examples": 30913, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46985963, "num_examples": 11460, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47418902, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 740992325, "size_in_bytes": 3116482868}, "id_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 386150473, "num_examples": 124866, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 53825037, "num_examples": 17144, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 105282698, "num_examples": 34300, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37985941, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37017309, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 620261458, "size_in_bytes": 2995752001}, "id_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 306605684, "num_examples": 85214, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 43990243, "num_examples": 11924, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86092519, "num_examples": 24014, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38763135, "num_examples": 10177, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44334776, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 519786357, "size_in_bytes": 2895276900}, "id_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 409187064, "num_examples": 142750, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 57347131, "num_examples": 19772, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 112292342, "num_examples": 39516, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35724755, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34646197, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 649197489, "size_in_bytes": 3024688032}, "id_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 293922032, "num_examples": 99711, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42650640, "num_examples": 14241, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 83306612, "num_examples": 28491, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27202479, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26320494, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 473402257, "size_in_bytes": 2848892800}, "id_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 526525523, "num_examples": 173971, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 75468199, "num_examples": 24669, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 148359222, "num_examples": 49239, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37200665, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36477348, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 824030957, "size_in_bytes": 3199521500}, "ar_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 133555405, "num_examples": 33580, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 19595585, "num_examples": 4761, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 38126567, "num_examples": 9517, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19671761, "num_examples": 4761, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26141703, "num_examples": 6676, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 237091021, "size_in_bytes": 2612581564}, "ar_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 191212853, "num_examples": 37609, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27773477, "num_examples": 5302, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54062298, "num_examples": 10671, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27858309, "num_examples": 5302, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42095898, "num_examples": 7830, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 343002835, "size_in_bytes": 2718493378}, "ar_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166948680, "num_examples": 42488, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23640937, "num_examples": 5913, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46790553, "num_examples": 11919, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23735545, "num_examples": 5913, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34842571, "num_examples": 9078, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 295958286, "size_in_bytes": 2671448829}, "ar_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 176483837, "num_examples": 51362, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25539348, "num_examples": 7225, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49710549, "num_examples": 14436, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25654948, "num_examples": 7225, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35958525, "num_examples": 10820, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 313347207, "size_in_bytes": 2688837750}, "ar_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169182989, "num_examples": 42469, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24388096, "num_examples": 5993, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47723955, "num_examples": 11945, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24483984, "num_examples": 5993, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35777939, "num_examples": 9104, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 301556963, "size_in_bytes": 2677047506}, "ar_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 241944775, "num_examples": 47948, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35190181, "num_examples": 6698, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 68345539, "num_examples": 13537, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35297349, "num_examples": 6698, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 56424995, "num_examples": 10696, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 437202839, "size_in_bytes": 2812693382}, "ar_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 368679326, "num_examples": 100111, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 53272378, "num_examples": 14189, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 104133848, "num_examples": 28172, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42324853, "num_examples": 10825, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46594170, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 615004575, "size_in_bytes": 2990495118}, "ar_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 263048332, "num_examples": 73181, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37565756, "num_examples": 10176, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73717887, "num_examples": 20483, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37313185, "num_examples": 10053, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43779436, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 455424596, "size_in_bytes": 2830915139}, "ar_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 110602032, "num_examples": 27899, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15960618, "num_examples": 3944, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31408171, "num_examples": 7941, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16023722, "num_examples": 3944, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19398091, "num_examples": 5100, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 193392634, "size_in_bytes": 2568883177}, "ar_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 231959596, "num_examples": 56694, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33705379, "num_examples": 7956, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65600585, "num_examples": 15962, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33832675, "num_examples": 7956, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49509101, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 414607336, "size_in_bytes": 2790097879}, "ar_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 289560780, "num_examples": 82842, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41015302, "num_examples": 11496, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80623597, "num_examples": 23118, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35048299, "num_examples": 9403, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44008429, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 490256407, "size_in_bytes": 2865746950}, "ar_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 396054199, "num_examples": 84835, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56567853, "num_examples": 11857, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 110756593, "num_examples": 23719, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 45488638, "num_examples": 9569, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 56334575, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 665201858, "size_in_bytes": 3040692401}, "ar_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 332912886, "num_examples": 93020, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47629352, "num_examples": 13056, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 93632278, "num_examples": 26144, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39055396, "num_examples": 10223, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45226891, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 558456803, "size_in_bytes": 2933947346}, "ar_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 306605684, "num_examples": 85214, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 43990243, "num_examples": 11924, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86092519, "num_examples": 24014, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38763135, "num_examples": 10177, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44334776, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 519786357, "size_in_bytes": 2895276900}, "ar_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 358513202, "num_examples": 110546, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 51110581, "num_examples": 15554, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 100978009, "num_examples": 31156, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37389558, "num_examples": 10389, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42783527, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 590774877, "size_in_bytes": 2966265420}, "ar_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 232200686, "num_examples": 61323, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33974274, "num_examples": 8757, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 66718590, "num_examples": 17523, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34114386, "num_examples": 8757, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34826534, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 401834470, "size_in_bytes": 2777325013}, "ar_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 464163271, "num_examples": 137477, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 67308522, "num_examples": 19675, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 132377972, "num_examples": 39109, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42786950, "num_examples": 11359, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44797137, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 751433852, "size_in_bytes": 3126924395}, "pt_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 194291996, "num_examples": 70642, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28028213, "num_examples": 10065, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55164572, "num_examples": 20159, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14219366, "num_examples": 4900, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19649329, "num_examples": 6828, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 311353476, "size_in_bytes": 2686844019}, "pt_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 255707310, "num_examples": 75571, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36885455, "num_examples": 10790, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 72039266, "num_examples": 21519, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23088208, "num_examples": 5625, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36545783, "num_examples": 8188, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 424266022, "size_in_bytes": 2799756565}, "pt_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 228533279, "num_examples": 80952, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32636435, "num_examples": 11507, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64262307, "num_examples": 23013, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18850660, "num_examples": 6342, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28791178, "num_examples": 9682, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 373073859, "size_in_bytes": 2748564402}, "pt_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 239533033, "num_examples": 91918, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34317307, "num_examples": 12997, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67387617, "num_examples": 26008, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20555372, "num_examples": 7832, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30120094, "num_examples": 11902, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 391913423, "size_in_bytes": 2767403966}, "pt_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 225683803, "num_examples": 79791, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32355907, "num_examples": 11353, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63767142, "num_examples": 22727, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18567668, "num_examples": 6188, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28292987, "num_examples": 9396, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 368667507, "size_in_bytes": 2744158050}, "pt_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 294294528, "num_examples": 84730, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42635895, "num_examples": 11994, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 82965864, "num_examples": 24067, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28857912, "num_examples": 6829, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47513149, "num_examples": 10736, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 496267348, "size_in_bytes": 2871757891}, "pt_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 515882894, "num_examples": 173551, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 68794446, "num_examples": 22831, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 134550579, "num_examples": 45266, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36848981, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36126524, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 792203424, "size_in_bytes": 3167693967}, "pt_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 320040173, "num_examples": 114523, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45452674, "num_examples": 15998, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89143645, "num_examples": 32097, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31323368, "num_examples": 10710, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33925525, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 519885385, "size_in_bytes": 2895375928}, "pt_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 176432382, "num_examples": 65719, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25253786, "num_examples": 9376, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49706176, "num_examples": 18765, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11433915, "num_examples": 4211, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14168629, "num_examples": 5434, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 276994888, "size_in_bytes": 2652485431}, "pt_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 279114345, "num_examples": 93098, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40260120, "num_examples": 13174, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78808326, "num_examples": 26370, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26501017, "num_examples": 8009, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39569743, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 464253551, "size_in_bytes": 2839744094}, "pt_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 426623553, "num_examples": 152968, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59201541, "num_examples": 21130, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 116270099, "num_examples": 42252, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34239466, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32997725, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 669332384, "size_in_bytes": 3044822927}, "pt_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 566144695, "num_examples": 154721, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 77493111, "num_examples": 20973, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 152245651, "num_examples": 42067, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46579271, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45276455, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 887739183, "size_in_bytes": 3263229726}, "pt_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 474936699, "num_examples": 165092, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64418438, "num_examples": 22196, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 126646414, "num_examples": 44278, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35609586, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34748803, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 736359940, "size_in_bytes": 3111850483}, "pt_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 409187064, "num_examples": 142750, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 57347131, "num_examples": 19772, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 112292342, "num_examples": 39516, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35724755, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34646197, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 649197489, "size_in_bytes": 3024688032}, "pt_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 358513202, "num_examples": 110546, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 51110581, "num_examples": 15554, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 100978009, "num_examples": 31156, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37389558, "num_examples": 10389, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42783527, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 590774877, "size_in_bytes": 2966265420}, "pt_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 456470296, "num_examples": 171477, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 65437965, "num_examples": 24495, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129182566, "num_examples": 48993, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24286741, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23389901, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 698767469, "size_in_bytes": 3074258012}, "pt_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 696823654, "num_examples": 245347, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 98646459, "num_examples": 34565, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 182590574, "num_examples": 64323, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34507564, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33983219, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 1046551470, "size_in_bytes": 3422042013}, "en_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 45397096, "num_examples": 15099, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6794491, "num_examples": 2154, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13085046, "num_examples": 4314, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6828955, "num_examples": 2154, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13154070, "num_examples": 4314, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 85259658, "size_in_bytes": 2460750201}, "en_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 112897266, "num_examples": 20826, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16397760, "num_examples": 2973, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31931211, "num_examples": 5952, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16445328, "num_examples": 2973, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32026443, "num_examples": 5952, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 209698008, "size_in_bytes": 2585188551}, "en_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 82825334, "num_examples": 26559, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11845377, "num_examples": 3792, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23220805, "num_examples": 7590, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11906049, "num_examples": 3792, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23340695, "num_examples": 7590, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 153138260, "size_in_bytes": 2528628803}, "en_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 96675723, "num_examples": 39633, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 14153668, "num_examples": 5658, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 27544132, "num_examples": 11325, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14244196, "num_examples": 5658, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21896965, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 174514684, "size_in_bytes": 2550005227}, "en_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81014576, "num_examples": 25572, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11682885, "num_examples": 3648, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22944915, "num_examples": 7308, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11741253, "num_examples": 3648, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23061843, "num_examples": 7308, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 150445472, "size_in_bytes": 2525936015}, "en_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 155128871, "num_examples": 30975, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23124256, "num_examples": 4425, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 43977504, "num_examples": 8850, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23195056, "num_examples": 4425, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44119104, "num_examples": 8850, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 289544791, "size_in_bytes": 2665035334}, "en_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 411286458, "num_examples": 133668, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59349317, "num_examples": 19092, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 117512924, "num_examples": 38193, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28068752, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28260502, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 644477953, "size_in_bytes": 3019968496}, "en_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 188073112, "num_examples": 65598, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27612857, "num_examples": 9369, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53427535, "num_examples": 18744, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26562758, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25844976, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 321521238, "size_in_bytes": 2697011781}, "en_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 25008753, "num_examples": 9444, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 3560378, "num_examples": 1347, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 6853199, "num_examples": 2700, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 3581930, "num_examples": 1347, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 6896399, "num_examples": 2700, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 45900659, "size_in_bytes": 2421391202}, "en_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 142275998, "num_examples": 41121, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21099930, "num_examples": 5871, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40754049, "num_examples": 11751, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21193866, "num_examples": 5871, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31471022, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 256794865, "size_in_bytes": 2632285408}, "en_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 300597006, "num_examples": 106983, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 43061023, "num_examples": 15279, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 84501587, "num_examples": 30567, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25513996, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24859349, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 478532961, "size_in_bytes": 2854023504}, "en_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 459684728, "num_examples": 111084, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 66397480, "num_examples": 15864, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129667377, "num_examples": 31740, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37928230, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36795657, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 730473472, "size_in_bytes": 3105964015}, "en_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 360916674, "num_examples": 122517, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 52048916, "num_examples": 17499, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 102578311, "num_examples": 35007, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27172663, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26585059, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 569301623, "size_in_bytes": 2944792166}, "en_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 293922032, "num_examples": 99711, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42650640, "num_examples": 14241, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 83306612, "num_examples": 28491, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27202479, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26320494, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 473402257, "size_in_bytes": 2848892800}, "en_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 232200686, "num_examples": 61323, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33974274, "num_examples": 8757, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 66718590, "num_examples": 17523, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34114386, "num_examples": 8757, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34826534, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 401834470, "size_in_bytes": 2777325013}, "en_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 456470296, "num_examples": 171477, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 65437965, "num_examples": 24495, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129182566, "num_examples": 48993, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24286741, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23389901, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 698767469, "size_in_bytes": 3074258012}, "en_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 678143650, "num_examples": 237636, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 96926929, "num_examples": 33948, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 193834997, "num_examples": 67896, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25904859, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25739390, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 1020549825, "size_in_bytes": 3396040368}, "es_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 280766481, "num_examples": 93787, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40430724, "num_examples": 13388, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80243228, "num_examples": 26766, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15808560, "num_examples": 5072, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21994928, "num_examples": 7134, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 439243921, "size_in_bytes": 2814734464}, "es_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 344440176, "num_examples": 98666, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 49718675, "num_examples": 14125, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 97646232, "num_examples": 28108, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25108303, "num_examples": 5809, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39419404, "num_examples": 8476, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 556332790, "size_in_bytes": 2931823333}, "es_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 316436712, "num_examples": 104203, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45292984, "num_examples": 14898, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89785881, "num_examples": 29710, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20694980, "num_examples": 6582, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31583135, "num_examples": 10078, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 503793692, "size_in_bytes": 2879284235}, "es_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 331493992, "num_examples": 116847, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 47691542, "num_examples": 16692, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 94514334, "num_examples": 33345, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23122242, "num_examples": 8376, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32092411, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 528914521, "size_in_bytes": 2904405064}, "es_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 315248680, "num_examples": 103456, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 45358161, "num_examples": 14808, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89824786, "num_examples": 29528, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20758717, "num_examples": 6492, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31620678, "num_examples": 9896, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 502811022, "size_in_bytes": 2878301565}, "es_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 387507666, "num_examples": 108637, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56359166, "num_examples": 15523, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 109985100, "num_examples": 30940, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31771162, "num_examples": 7207, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51803584, "num_examples": 11308, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 637426678, "size_in_bytes": 3012917221}, "es_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 644793465, "num_examples": 207080, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 91017234, "num_examples": 28904, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 179593516, "num_examples": 57533, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38444271, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38352312, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 992200798, "size_in_bytes": 3367691341}, "es_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 422075349, "num_examples": 141670, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 60963067, "num_examples": 20183, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 119689771, "num_examples": 40212, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36034236, "num_examples": 11744, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35983844, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 674746267, "size_in_bytes": 3050236810}, "es_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 259722926, "num_examples": 87986, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37191372, "num_examples": 12591, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73811477, "num_examples": 25112, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12556456, "num_examples": 4275, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15536713, "num_examples": 5480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 398818944, "size_in_bytes": 2774309487}, "es_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 375252849, "num_examples": 118093, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 54466998, "num_examples": 16897, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 106888168, "num_examples": 33617, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29900978, "num_examples": 8581, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 41566919, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 608075912, "size_in_bytes": 2983566455}, "es_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 536239898, "num_examples": 181897, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 76139890, "num_examples": 25749, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 150448276, "num_examples": 51463, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35603354, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35018317, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 833449735, "size_in_bytes": 3208940278}, "es_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 685054819, "num_examples": 184296, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 97093934, "num_examples": 25888, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 190739233, "num_examples": 51730, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 48206822, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47058922, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 1068153730, "size_in_bytes": 3443644273}, "es_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 595260381, "num_examples": 196565, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 84007562, "num_examples": 27557, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 166264097, "num_examples": 54989, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37244632, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36783845, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 919560517, "size_in_bytes": 3295051060}, "es_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 526525523, "num_examples": 173971, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 75468199, "num_examples": 24669, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 148359222, "num_examples": 49239, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37200665, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36477348, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 824030957, "size_in_bytes": 3199521500}, "es_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 464163271, "num_examples": 137477, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 67308522, "num_examples": 19675, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 132377972, "num_examples": 39109, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42786950, "num_examples": 11359, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44797137, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 751433852, "size_in_bytes": 3126924395}, "es_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 696823654, "num_examples": 245347, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 98646459, "num_examples": 34565, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 182590574, "num_examples": 64323, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34507564, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33983219, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 1046551470, "size_in_bytes": 3422042013}, "es_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 678143650, "num_examples": 237636, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 96926929, "num_examples": 33948, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 193834997, "num_examples": 67896, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25904859, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25739390, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 1020549825, "size_in_bytes": 3396040368}, "multilingual": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "multilingual", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1732045467, "num_examples": 538762, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 249193591, "num_examples": 76627, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 492718075, "num_examples": 154262, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 140135756, "num_examples": 39875, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 171173577, "num_examples": 48238, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 2785266466, "size_in_bytes": 5160757009}, "crosslingual": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "crosslingual", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 13142805739, "num_examples": 3884868, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 1816841565, "num_examples": 524500, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 3510998462, "num_examples": 1042862, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 1540229811, "num_examples": 430574, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 2245084473, "num_examples": 634520, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_GEM_v2.tar.gz": {"num_bytes": 2375490543, "checksum": "340ae286e674c4c138f5412a14694e10130d8e7f4722feefe0987d71c7918a1c"}}, "download_size": 2375490543, "post_processing_size": null, "dataset_size": 22255960050, "size_in_bytes": 24631450593}} \ No newline at end of file +{"cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 15002564, "num_examples": 4872, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 2225539, "num_examples": 694, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 4305260, "num_examples": 1388, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 2236643, "num_examples": 694, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 4327468, "num_examples": 1388, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 28097474, "size_in_bytes": 2363420868}, "hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 46507451, "num_examples": 6724, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6755548, "num_examples": 961, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13146740, "num_examples": 1934, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6770924, "num_examples": 961, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13177684, "num_examples": 1934, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 86358347, "size_in_bytes": 2421681741}, "ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 29352806, "num_examples": 8657, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4166865, "num_examples": 1226, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 8218287, "num_examples": 2480, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4186481, "num_examples": 1226, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8256799, "num_examples": 2480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 54181238, "size_in_bytes": 2389504632}, "zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 29268617, "num_examples": 12523, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4301955, "num_examples": 1797, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 8330353, "num_examples": 3584, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4330707, "num_examples": 1797, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 6985757, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 53217389, "size_in_bytes": 2388540783}, "ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 28032747, "num_examples": 8370, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 4032251, "num_examples": 1188, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 7915245, "num_examples": 2396, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 4051259, "num_examples": 1188, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7953581, "num_examples": 2396, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 51985083, "size_in_bytes": 2387308477}, "th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 62897852, "num_examples": 10058, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 9368369, "num_examples": 1437, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 17790568, "num_examples": 2876, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9391361, "num_examples": 1437, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 17836584, "num_examples": 2876, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 117284734, "size_in_bytes": 2452608128}, "fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 143223283, "num_examples": 43423, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20643946, "num_examples": 6193, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40910697, "num_examples": 12405, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10046850, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 9920190, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 224744966, "size_in_bytes": 2560068360}, "nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 63569060, "num_examples": 21345, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 9333929, "num_examples": 3058, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 18042359, "num_examples": 6105, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9213907, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8887843, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 109047098, "size_in_bytes": 2444370492}, "tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 8292548, "num_examples": 3052, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 1184767, "num_examples": 438, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 2261072, "num_examples": 874, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 1191775, "num_examples": 438, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 2275056, "num_examples": 874, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 15205218, "size_in_bytes": 2350528612}, "vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 50413922, "num_examples": 13262, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7479135, "num_examples": 1892, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 14420574, "num_examples": 3789, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7509407, "num_examples": 1892, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 11478313, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 91301351, "size_in_bytes": 2426624745}, "it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 100608731, "num_examples": 34085, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 14352177, "num_examples": 4850, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 28098290, "num_examples": 9643, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8885968, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8744337, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 160689503, "size_in_bytes": 2496012897}, "ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 175965263, "num_examples": 35313, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25250672, "num_examples": 4984, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49241490, "num_examples": 9962, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15221365, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14822702, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 280501492, "size_in_bytes": 2615824886}, "de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 123151571, "num_examples": 39505, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17760316, "num_examples": 5643, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 34952685, "num_examples": 11266, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9486989, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 9339041, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 194690602, "size_in_bytes": 2530013996}, "id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 99448359, "num_examples": 32228, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 14366904, "num_examples": 4596, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 28037353, "num_examples": 9183, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9392366, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 9280503, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 160525485, "size_in_bytes": 2495848879}, "ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 86368475, "num_examples": 19992, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12641361, "num_examples": 2859, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24802765, "num_examples": 5710, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12687105, "num_examples": 2859, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13034139, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 149533845, "size_in_bytes": 2484857239}, "pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 148803666, "num_examples": 54673, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21257644, "num_examples": 7782, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 41893060, "num_examples": 15513, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8236034, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 8102489, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 228292893, "size_in_bytes": 2563616287}, "en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 251472856, "num_examples": 95517, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35102072, "num_examples": 13340, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 71957076, "num_examples": 27489, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7949889, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7901255, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 374383148, "size_in_bytes": 2709706542}, "es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 229654469, "num_examples": 76295, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32829970, "num_examples": 10903, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65477611, "num_examples": 21726, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 9020574, "num_examples": 3000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 9234806, "num_examples": 3000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 346217430, "size_in_bytes": 2681540824}, "cs_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81528693, "num_examples": 15454, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11565497, "num_examples": 2129, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22650770, "num_examples": 4324, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11599561, "num_examples": 2129, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22719954, "num_examples": 4324, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 150064475, "size_in_bytes": 2485387869}, "cs_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 57445863, "num_examples": 17465, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8217462, "num_examples": 2438, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 15842425, "num_examples": 4820, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8256470, "num_examples": 2438, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15918377, "num_examples": 4820, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 105680597, "size_in_bytes": 2441003991}, "cs_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 59778990, "num_examples": 22981, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8655747, "num_examples": 3229, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16769247, "num_examples": 6444, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8707411, "num_examples": 3229, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15470411, "num_examples": 5860, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 109381806, "size_in_bytes": 2444705200}, "cs_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 59011682, "num_examples": 18012, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8329385, "num_examples": 2492, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16220831, "num_examples": 4954, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8369257, "num_examples": 2492, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16300095, "num_examples": 4954, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 108231250, "size_in_bytes": 2443554644}, "cs_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 106621112, "num_examples": 20734, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15535648, "num_examples": 2871, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29989332, "num_examples": 5810, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15581584, "num_examples": 2871, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30082292, "num_examples": 5810, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 197809968, "size_in_bytes": 2533133362}, "cs_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 186363152, "num_examples": 56741, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26941961, "num_examples": 8065, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52886769, "num_examples": 16097, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16374817, "num_examples": 4872, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21955334, "num_examples": 6692, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 304522033, "size_in_bytes": 2639845427}, "cs_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 103521326, "num_examples": 34191, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15092589, "num_examples": 4826, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29371845, "num_examples": 9709, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15000855, "num_examples": 4768, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20274993, "num_examples": 6604, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 183261608, "size_in_bytes": 2518585002}, "cs_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 27098573, "num_examples": 9234, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 3894999, "num_examples": 1290, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 7353546, "num_examples": 2550, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 3915639, "num_examples": 1290, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7394346, "num_examples": 2550, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 49657103, "size_in_bytes": 2384980497}, "cs_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 88462610, "num_examples": 24638, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13009081, "num_examples": 3468, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24918723, "num_examples": 6909, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13064569, "num_examples": 3468, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22026382, "num_examples": 6120, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 161481365, "size_in_bytes": 2496804759}, "cs_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 139415715, "num_examples": 46519, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20028591, "num_examples": 6592, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39070072, "num_examples": 13123, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14590254, "num_examples": 4742, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19771799, "num_examples": 6480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 232876431, "size_in_bytes": 2568199825}, "cs_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 222796210, "num_examples": 47777, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32376186, "num_examples": 6780, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62643370, "num_examples": 13486, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22375615, "num_examples": 4796, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28280966, "num_examples": 6524, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 368472347, "size_in_bytes": 2703795741}, "cs_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 164086715, "num_examples": 52293, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23764268, "num_examples": 7431, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46438098, "num_examples": 14834, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15519549, "num_examples": 4788, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20881542, "num_examples": 6568, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 270690172, "size_in_bytes": 2606013566}, "cs_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 140142281, "num_examples": 45146, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20241248, "num_examples": 6382, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39456250, "num_examples": 12801, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15295286, "num_examples": 4786, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20757288, "num_examples": 6618, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 235892353, "size_in_bytes": 2571215747}, "cs_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 131413705, "num_examples": 32728, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 19265725, "num_examples": 4649, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 37437193, "num_examples": 9266, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19340109, "num_examples": 4649, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25725463, "num_examples": 6556, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 233182195, "size_in_bytes": 2568505589}, "cs_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 188786420, "num_examples": 67687, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27128353, "num_examples": 9624, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53313973, "num_examples": 19197, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14136215, "num_examples": 4842, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19582346, "num_examples": 6684, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 302947307, "size_in_bytes": 2638270701}, "cs_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 44453638, "num_examples": 14616, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6612386, "num_examples": 2082, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 12756303, "num_examples": 4164, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6645698, "num_examples": 2082, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 12822927, "num_examples": 4164, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 83290952, "size_in_bytes": 2418614346}, "cs_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "cs_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 274283144, "num_examples": 90407, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 39462727, "num_examples": 12911, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78171714, "num_examples": 25720, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15685459, "num_examples": 5008, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21992813, "num_examples": 6994, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 429595857, "size_in_bytes": 2764919251}, "hi_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81528693, "num_examples": 15454, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11565497, "num_examples": 2129, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22650770, "num_examples": 4324, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11599561, "num_examples": 2129, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22719954, "num_examples": 4324, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 150064475, "size_in_bytes": 2485387869}, "hi_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 118290783, "num_examples": 23653, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16714405, "num_examples": 3331, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 32546027, "num_examples": 6664, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16767701, "num_examples": 3331, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32651483, "num_examples": 6664, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 216970399, "size_in_bytes": 2552293793}, "hi_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 108732957, "num_examples": 26439, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15658257, "num_examples": 3724, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29918634, "num_examples": 7448, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15717841, "num_examples": 3724, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28635862, "num_examples": 6864, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 198663551, "size_in_bytes": 2533986945}, "hi_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 112983533, "num_examples": 22652, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15588273, "num_examples": 3097, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30947550, "num_examples": 6326, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15637825, "num_examples": 3097, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31048766, "num_examples": 6326, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 206205947, "size_in_bytes": 2541529341}, "hi_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166168643, "num_examples": 25642, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23963204, "num_examples": 3558, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 45992560, "num_examples": 7226, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24020132, "num_examples": 3558, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46108176, "num_examples": 7226, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 306252715, "size_in_bytes": 2641576109}, "hi_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 246278795, "num_examples": 61173, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35511650, "num_examples": 8710, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69644583, "num_examples": 17437, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24954826, "num_examples": 5517, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38734588, "num_examples": 8032, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 415124442, "size_in_bytes": 2750447836}, "hi_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 153783851, "num_examples": 36983, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22260690, "num_examples": 5229, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 43342373, "num_examples": 10555, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22175404, "num_examples": 5171, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34259057, "num_examples": 7450, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 275821375, "size_in_bytes": 2611144769}, "hi_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 73498845, "num_examples": 13818, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10404433, "num_examples": 1945, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20388674, "num_examples": 3954, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10435553, "num_examples": 1945, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20451938, "num_examples": 3954, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 135179443, "size_in_bytes": 2470502837}, "hi_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 148025719, "num_examples": 29556, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21731708, "num_examples": 4193, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 41452971, "num_examples": 8409, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21798796, "num_examples": 4193, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38584630, "num_examples": 7620, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 271593824, "size_in_bytes": 2606917218}, "hi_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 192953917, "num_examples": 49985, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27761700, "num_examples": 7103, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53523513, "num_examples": 14077, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22331539, "num_examples": 5253, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34240504, "num_examples": 7434, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 330811173, "size_in_bytes": 2666134567}, "hi_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 279329223, "num_examples": 51611, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40283523, "num_examples": 7303, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 77675220, "num_examples": 14536, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 30291320, "num_examples": 5319, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43329616, "num_examples": 7574, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 470908902, "size_in_bytes": 2806232296}, "hi_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221028649, "num_examples": 56409, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32022615, "num_examples": 8040, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62270935, "num_examples": 16080, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23787640, "num_examples": 5397, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36734315, "num_examples": 7814, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 375844154, "size_in_bytes": 2711167548}, "hi_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 195370301, "num_examples": 48962, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27987737, "num_examples": 6903, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54356009, "num_examples": 13905, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23050111, "num_examples": 5307, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35674711, "num_examples": 7722, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 336438869, "size_in_bytes": 2671762263}, "hi_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 187631967, "num_examples": 36518, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27287921, "num_examples": 5162, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52992331, "num_examples": 10366, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27370513, "num_examples": 5162, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 41298201, "num_examples": 7656, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 336580933, "size_in_bytes": 2671904327}, "hi_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 247952847, "num_examples": 72147, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35719383, "num_examples": 10293, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69647118, "num_examples": 20479, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22737949, "num_examples": 5511, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35936003, "num_examples": 7966, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 411993300, "size_in_bytes": 2747316694}, "hi_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 110618152, "num_examples": 20172, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16073147, "num_examples": 2883, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31274304, "num_examples": 5802, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16119275, "num_examples": 2883, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31367136, "num_examples": 5802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 205452014, "size_in_bytes": 2540775408}, "hi_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "hi_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 336530602, "num_examples": 95119, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 48593955, "num_examples": 13628, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 95243183, "num_examples": 27064, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24828159, "num_examples": 5725, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39085786, "num_examples": 8338, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 544281685, "size_in_bytes": 2879605079}, "ja_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 57445863, "num_examples": 17465, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8217462, "num_examples": 2438, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 15842425, "num_examples": 4820, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8256470, "num_examples": 2438, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15918377, "num_examples": 4820, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 105680597, "size_in_bytes": 2441003991}, "ja_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 118290783, "num_examples": 23653, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16714405, "num_examples": 3331, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 32546027, "num_examples": 6664, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16767701, "num_examples": 3331, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32651483, "num_examples": 6664, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 216970399, "size_in_bytes": 2552293793}, "ja_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 85081957, "num_examples": 30570, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11966134, "num_examples": 4225, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23346286, "num_examples": 8522, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12033734, "num_examples": 4225, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22079530, "num_examples": 7938, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 154507641, "size_in_bytes": 2489831035}, "ja_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 87250166, "num_examples": 26083, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12045182, "num_examples": 3596, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23848616, "num_examples": 7252, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12102718, "num_examples": 3596, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23963480, "num_examples": 7252, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 159210162, "size_in_bytes": 2494533556}, "ja_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 141393601, "num_examples": 29101, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 19929165, "num_examples": 3989, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 38342290, "num_examples": 8016, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19992989, "num_examples": 3989, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38469378, "num_examples": 8016, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 258127423, "size_in_bytes": 2593450817}, "ja_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221099189, "num_examples": 66684, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31632274, "num_examples": 9477, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62386441, "num_examples": 18975, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21087722, "num_examples": 6284, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31499886, "num_examples": 9570, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 367705512, "size_in_bytes": 2703028906}, "ja_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 130602778, "num_examples": 41988, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18575134, "num_examples": 5866, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36296549, "num_examples": 11851, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18500040, "num_examples": 5808, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27232801, "num_examples": 8746, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 231207302, "size_in_bytes": 2566530696}, "ja_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 50083569, "num_examples": 15861, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7032401, "num_examples": 2248, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13702927, "num_examples": 4500, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7068369, "num_examples": 2248, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13773759, "num_examples": 4500, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 91661025, "size_in_bytes": 2426984419}, "ja_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 122373588, "num_examples": 34039, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17415440, "num_examples": 4726, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33696272, "num_examples": 9493, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17491056, "num_examples": 4726, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30844107, "num_examples": 8704, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 221820463, "size_in_bytes": 2557143857}, "ja_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 167429228, "num_examples": 54520, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23647408, "num_examples": 7676, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46393975, "num_examples": 15377, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18226415, "num_examples": 5826, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27130598, "num_examples": 8734, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 282827624, "size_in_bytes": 2618151018}, "ja_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 256536185, "num_examples": 56270, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36624583, "num_examples": 7914, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 71046411, "num_examples": 15814, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26642156, "num_examples": 5930, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36720087, "num_examples": 8852, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 427569422, "size_in_bytes": 2762892816}, "ja_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 196039909, "num_examples": 61638, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28068370, "num_examples": 8725, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55190893, "num_examples": 17542, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19844355, "num_examples": 6082, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29676497, "num_examples": 9276, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 328820024, "size_in_bytes": 2664143418}, "ja_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169716742, "num_examples": 53785, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23942457, "num_examples": 7514, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47145159, "num_examples": 15213, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19014607, "num_examples": 5918, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28483621, "num_examples": 9030, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 288302586, "size_in_bytes": 2623625980}, "ja_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 164263852, "num_examples": 41471, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23242384, "num_examples": 5771, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 45970025, "num_examples": 11638, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23334720, "num_examples": 5771, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34295079, "num_examples": 8928, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 291106060, "size_in_bytes": 2626429454}, "ja_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221794282, "num_examples": 77592, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31524351, "num_examples": 10986, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62126672, "num_examples": 21977, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18554005, "num_examples": 6204, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28438357, "num_examples": 9464, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 362437667, "size_in_bytes": 2697761061}, "ja_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81409319, "num_examples": 25971, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11560905, "num_examples": 3678, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22773109, "num_examples": 7440, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11619753, "num_examples": 3678, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22890981, "num_examples": 7440, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 150254067, "size_in_bytes": 2485577461}, "ja_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ja_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 309354143, "num_examples": 100740, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44193203, "num_examples": 14375, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 87558590, "num_examples": 28660, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20439359, "num_examples": 6472, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31425561, "num_examples": 9934, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 492970856, "size_in_bytes": 2828294250}, "zh_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 59778990, "num_examples": 22981, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8655747, "num_examples": 3229, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16769247, "num_examples": 6444, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8707411, "num_examples": 3229, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15470411, "num_examples": 5860, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 109381806, "size_in_bytes": 2444705200}, "zh_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 108732957, "num_examples": 26439, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15658257, "num_examples": 3724, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29918634, "num_examples": 7448, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15717841, "num_examples": 3724, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28635862, "num_examples": 6864, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 198663551, "size_in_bytes": 2533986945}, "zh_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 85081957, "num_examples": 30570, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11966134, "num_examples": 4225, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23346286, "num_examples": 8522, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12033734, "num_examples": 4225, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22079530, "num_examples": 7938, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 154507641, "size_in_bytes": 2489831035}, "zh_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 80910264, "num_examples": 29375, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11500432, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22134330, "num_examples": 8108, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11565632, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20862118, "num_examples": 7524, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 146972776, "size_in_bytes": 2482296170}, "zh_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 141205589, "num_examples": 34319, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20416447, "num_examples": 4748, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 38691157, "num_examples": 9500, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20492415, "num_examples": 4748, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37441217, "num_examples": 8916, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 258246825, "size_in_bytes": 2593570219}, "zh_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 231879240, "num_examples": 76962, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33584762, "num_examples": 10952, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65496953, "num_examples": 21765, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23063810, "num_examples": 7759, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33254266, "num_examples": 11776, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 387279031, "size_in_bytes": 2722602425}, "zh_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 140654490, "num_examples": 51756, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20385605, "num_examples": 7287, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39360755, "num_examples": 14563, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20333247, "num_examples": 7229, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28939627, "num_examples": 10874, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 249673724, "size_in_bytes": 2584997118}, "zh_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 45248381, "num_examples": 18769, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6527589, "num_examples": 2665, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 12524628, "num_examples": 5312, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6570229, "num_examples": 2665, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 11207680, "num_examples": 4728, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 82078507, "size_in_bytes": 2417401901}, "zh_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 120875160, "num_examples": 39555, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17565187, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33254055, "num_examples": 10909, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17653427, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29023774, "num_examples": 9536, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 218371603, "size_in_bytes": 2553694997}, "zh_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 177916880, "num_examples": 64406, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25411901, "num_examples": 9101, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49344464, "num_examples": 18055, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20013708, "num_examples": 7251, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28723163, "num_examples": 10828, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 301410116, "size_in_bytes": 2636733510}, "zh_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 271938701, "num_examples": 66166, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 39039165, "num_examples": 9317, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 75905428, "num_examples": 18614, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29079186, "num_examples": 7333, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40223132, "num_examples": 11068, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 456185612, "size_in_bytes": 2791509006}, "zh_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 207428120, "num_examples": 71914, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 29904137, "num_examples": 10184, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 58657522, "num_examples": 20396, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21703466, "num_examples": 7541, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31788018, "num_examples": 11546, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 349481263, "size_in_bytes": 2684804657}, "zh_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 180732752, "num_examples": 64011, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26075830, "num_examples": 9053, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 50607034, "num_examples": 18051, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21172604, "num_examples": 7457, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30590132, "num_examples": 11284, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 309178352, "size_in_bytes": 2644501746}, "zh_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 172246002, "num_examples": 49583, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24970206, "num_examples": 7004, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48502033, "num_examples": 13938, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25082270, "num_examples": 7004, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35463115, "num_examples": 10644, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 306263626, "size_in_bytes": 2641587020}, "zh_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 230787450, "num_examples": 87412, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33050829, "num_examples": 12369, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64700437, "num_examples": 24669, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20102611, "num_examples": 7587, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29654422, "num_examples": 11572, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 378295749, "size_in_bytes": 2713619143}, "zh_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 92487460, "num_examples": 37569, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13596249, "num_examples": 5391, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 26321891, "num_examples": 10752, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13682505, "num_examples": 5391, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22059198, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 168147303, "size_in_bytes": 2503470697}, "zh_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "zh_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 321867278, "num_examples": 112030, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46353595, "num_examples": 16036, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91572010, "num_examples": 31906, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22626327, "num_examples": 8133, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32425726, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 514844936, "size_in_bytes": 2850168330}, "ko_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 59011682, "num_examples": 18012, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 8329385, "num_examples": 2492, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 16220831, "num_examples": 4954, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 8369257, "num_examples": 2492, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16300095, "num_examples": 4954, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 108231250, "size_in_bytes": 2443554644}, "ko_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 112983533, "num_examples": 22652, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15588273, "num_examples": 3097, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30947550, "num_examples": 6326, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15637825, "num_examples": 3097, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31048766, "num_examples": 6326, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 206205947, "size_in_bytes": 2541529341}, "ko_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 87250166, "num_examples": 26083, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12045182, "num_examples": 3596, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 23848616, "num_examples": 7252, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12102718, "num_examples": 3596, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 23963480, "num_examples": 7252, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 159210162, "size_in_bytes": 2494533556}, "ko_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 80910264, "num_examples": 29375, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11500432, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22134330, "num_examples": 8108, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11565632, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20862118, "num_examples": 7524, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 146972776, "size_in_bytes": 2482296170}, "ko_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 145979982, "num_examples": 29670, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20679646, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39888495, "num_examples": 8236, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20744846, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40020271, "num_examples": 8236, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 267313240, "size_in_bytes": 2602636634}, "ko_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221148327, "num_examples": 66365, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31918853, "num_examples": 9453, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62653564, "num_examples": 18885, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21373917, "num_examples": 6260, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31766737, "num_examples": 9480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 368861398, "size_in_bytes": 2704184792}, "ko_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 129371176, "num_examples": 41391, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18440373, "num_examples": 5774, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36290986, "num_examples": 11691, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18363807, "num_examples": 5716, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27225846, "num_examples": 8586, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 229692188, "size_in_bytes": 2565015582}, "ko_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 48470658, "num_examples": 15778, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6912527, "num_examples": 2242, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13549391, "num_examples": 4520, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6948399, "num_examples": 2242, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13621711, "num_examples": 4520, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 89502686, "size_in_bytes": 2424826080}, "ko_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 126974933, "num_examples": 34898, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18308407, "num_examples": 4890, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 35737602, "num_examples": 9881, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18386647, "num_examples": 4890, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32892813, "num_examples": 9092, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 232300402, "size_in_bytes": 2567623796}, "ko_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166337969, "num_examples": 54061, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23603565, "num_examples": 7606, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46541821, "num_examples": 15269, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18181452, "num_examples": 5756, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27277884, "num_examples": 8626, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 281942691, "size_in_bytes": 2617266085}, "ko_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 254388387, "num_examples": 55519, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36467209, "num_examples": 7814, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 71219647, "num_examples": 15682, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26483182, "num_examples": 5830, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36892379, "num_examples": 8720, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 425450804, "size_in_bytes": 2760774198}, "ko_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 193559597, "num_examples": 60607, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27659067, "num_examples": 8557, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54581778, "num_examples": 17236, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19432364, "num_examples": 5914, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29063654, "num_examples": 8970, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 324296460, "size_in_bytes": 2659619854}, "ko_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 170279570, "num_examples": 53748, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24544740, "num_examples": 7592, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47749544, "num_examples": 15273, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19618138, "num_examples": 5996, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29090134, "num_examples": 9090, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 291282126, "size_in_bytes": 2626605520}, "ko_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166772243, "num_examples": 41564, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24045339, "num_examples": 5859, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46975608, "num_examples": 11680, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24139083, "num_examples": 5859, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35302502, "num_examples": 8970, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 297234775, "size_in_bytes": 2632558169}, "ko_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 219249985, "num_examples": 76567, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31333782, "num_examples": 10850, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 61684803, "num_examples": 21707, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18361260, "num_examples": 6068, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27993336, "num_examples": 9194, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 358623166, "size_in_bytes": 2693946560}, "ko_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 79823798, "num_examples": 25110, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11488843, "num_examples": 3564, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22564617, "num_examples": 7188, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11545867, "num_examples": 3564, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22679625, "num_examples": 7188, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 148102750, "size_in_bytes": 2483426144}, "ko_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ko_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 308342677, "num_examples": 100095, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44330902, "num_examples": 14315, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 87641207, "num_examples": 28502, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20576098, "num_examples": 6412, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31506818, "num_examples": 9776, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 492397702, "size_in_bytes": 2827721096}, "th_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 106621112, "num_examples": 20734, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15535648, "num_examples": 2871, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29989332, "num_examples": 5810, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15581584, "num_examples": 2871, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30082292, "num_examples": 5810, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 197809968, "size_in_bytes": 2533133362}, "th_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166168643, "num_examples": 25642, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23963204, "num_examples": 3558, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 45992560, "num_examples": 7226, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24020132, "num_examples": 3558, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46108176, "num_examples": 7226, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 306252715, "size_in_bytes": 2641576109}, "th_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 141393601, "num_examples": 29101, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 19929165, "num_examples": 3989, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 38342290, "num_examples": 8016, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19992989, "num_examples": 3989, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38469378, "num_examples": 8016, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 258127423, "size_in_bytes": 2593450817}, "th_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 141205589, "num_examples": 34319, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20416447, "num_examples": 4748, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 38691157, "num_examples": 9500, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20492415, "num_examples": 4748, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37441217, "num_examples": 8916, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 258246825, "size_in_bytes": 2593570219}, "th_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 145979982, "num_examples": 29670, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20679646, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39888495, "num_examples": 8236, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20744846, "num_examples": 4075, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40020271, "num_examples": 8236, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 267313240, "size_in_bytes": 2602636634}, "th_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 289309715, "num_examples": 70633, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42108191, "num_examples": 10034, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81471328, "num_examples": 20013, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31572551, "num_examples": 6841, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 50602549, "num_examples": 10608, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 495064334, "size_in_bytes": 2830387728}, "th_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 196188307, "num_examples": 46283, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28400579, "num_examples": 6443, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54501140, "num_examples": 12999, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28334717, "num_examples": 6385, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45456928, "num_examples": 9894, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 352881671, "size_in_bytes": 2688205065}, "th_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 87989691, "num_examples": 17314, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12878522, "num_examples": 2463, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24653361, "num_examples": 4970, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12917930, "num_examples": 2463, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24732881, "num_examples": 4970, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 163172385, "size_in_bytes": 2498495779}, "th_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 194630834, "num_examples": 39460, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28510169, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54262115, "num_examples": 11087, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28598409, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51436622, "num_examples": 10298, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 357438149, "size_in_bytes": 2692761543}, "th_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 229125770, "num_examples": 58235, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33012428, "num_examples": 8183, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63609914, "num_examples": 16337, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27599547, "num_examples": 6333, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44363065, "num_examples": 9694, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 397710724, "size_in_bytes": 2733034118}, "th_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 319843264, "num_examples": 59843, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46169619, "num_examples": 8373, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89184436, "num_examples": 16808, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36194536, "num_examples": 6389, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 54875184, "num_examples": 9846, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 546267039, "size_in_bytes": 2881590433}, "th_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 260545427, "num_examples": 65317, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37782101, "num_examples": 9220, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73264764, "num_examples": 18524, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29566006, "num_examples": 6577, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47767248, "num_examples": 10258, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 448925546, "size_in_bytes": 2784248940}, "th_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 236710901, "num_examples": 58252, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34194921, "num_examples": 8175, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65735448, "num_examples": 16439, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29277647, "num_examples": 6579, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47094694, "num_examples": 10256, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 413013611, "size_in_bytes": 2748337005}, "th_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 238218279, "num_examples": 46708, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34684531, "num_examples": 6538, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67280813, "num_examples": 13190, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34789139, "num_examples": 6538, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 55631867, "num_examples": 10480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 430604629, "size_in_bytes": 2765928023}, "th_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 286289876, "num_examples": 81039, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41347017, "num_examples": 11437, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80409979, "num_examples": 22905, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28383887, "num_examples": 6655, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46737680, "num_examples": 10392, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 483168439, "size_in_bytes": 2818491833}, "th_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 152727050, "num_examples": 30174, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22743105, "num_examples": 4311, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 43247373, "num_examples": 8628, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22812081, "num_examples": 4311, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43385421, "num_examples": 8628, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 284915030, "size_in_bytes": 2620238424}, "th_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "th_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 379460557, "num_examples": 104975, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55164438, "num_examples": 15010, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 107476969, "num_examples": 29824, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31420754, "num_examples": 7107, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51363732, "num_examples": 11098, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 624886450, "size_in_bytes": 2960209844}, "fr_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 186363152, "num_examples": 56741, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26941961, "num_examples": 8065, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52886769, "num_examples": 16097, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16374817, "num_examples": 4872, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21955334, "num_examples": 6692, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 304522033, "size_in_bytes": 2639845427}, "fr_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 246278795, "num_examples": 61173, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35511650, "num_examples": 8710, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69644583, "num_examples": 17437, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24954826, "num_examples": 5517, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38734588, "num_examples": 8032, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 415124442, "size_in_bytes": 2750447836}, "fr_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221099189, "num_examples": 66684, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31632274, "num_examples": 9477, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62386441, "num_examples": 18975, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21087722, "num_examples": 6284, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31499886, "num_examples": 9570, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 367705512, "size_in_bytes": 2703028906}, "fr_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 231879240, "num_examples": 76962, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33584762, "num_examples": 10952, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65496953, "num_examples": 21765, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23063810, "num_examples": 7759, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33254266, "num_examples": 11776, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 387279031, "size_in_bytes": 2722602425}, "fr_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221148327, "num_examples": 66365, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31918853, "num_examples": 9453, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62653564, "num_examples": 18885, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21373917, "num_examples": 6260, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31766737, "num_examples": 9480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 368861398, "size_in_bytes": 2704184792}, "fr_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 289309715, "num_examples": 70633, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 42108191, "num_examples": 10034, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81471328, "num_examples": 20013, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31572551, "num_examples": 6841, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 50602549, "num_examples": 10608, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 495064334, "size_in_bytes": 2830387728}, "fr_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 322869564, "num_examples": 101608, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46519904, "num_examples": 14341, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 90310307, "num_examples": 28514, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35884226, "num_examples": 11090, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37816296, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 533400297, "size_in_bytes": 2868723691}, "fr_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166313237, "num_examples": 51569, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24005949, "num_examples": 7379, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47270362, "num_examples": 14737, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13427829, "num_examples": 4186, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16317167, "num_examples": 5332, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 267334544, "size_in_bytes": 2602657938}, "fr_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 278142529, "num_examples": 79631, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40387383, "num_examples": 11283, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78432015, "num_examples": 22470, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29871727, "num_examples": 8090, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43587836, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 470421490, "size_in_bytes": 2805744884}, "fr_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 408425072, "num_examples": 129572, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55208131, "num_examples": 17277, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 108498129, "num_examples": 34496, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38427450, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37873247, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 648432029, "size_in_bytes": 2983755423}, "fr_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 544791012, "num_examples": 132550, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 72902906, "num_examples": 17431, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 141797425, "num_examples": 34639, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 51228388, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 50153085, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 860872816, "size_in_bytes": 3196196210}, "fr_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 478710842, "num_examples": 148610, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64658873, "num_examples": 19754, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 127179527, "num_examples": 39359, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39564815, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38692244, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 748806301, "size_in_bytes": 3084129695}, "fr_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 419393003, "num_examples": 129987, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59255562, "num_examples": 18095, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 116049787, "num_examples": 36136, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39446774, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38401219, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 672546345, "size_in_bytes": 3007869739}, "fr_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 363338130, "num_examples": 97839, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 52510566, "num_examples": 13880, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 102637261, "num_examples": 27553, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42036462, "num_examples": 10687, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46790994, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 607313413, "size_in_bytes": 2942636807}, "fr_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 503678107, "num_examples": 167390, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 66995531, "num_examples": 21979, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 131128053, "num_examples": 43578, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37166280, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36642991, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 775610962, "size_in_bytes": 3110934356}, "fr_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 404347028, "num_examples": 130269, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 58305760, "num_examples": 18579, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 115530811, "num_examples": 37215, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28395953, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28017229, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 634596781, "size_in_bytes": 2969920175}, "fr_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "fr_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 632129442, "num_examples": 200858, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 89191353, "num_examples": 28040, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 175918668, "num_examples": 55753, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38602577, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38440897, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 974282937, "size_in_bytes": 3309606331}, "nl_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 103521326, "num_examples": 34191, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15092589, "num_examples": 4826, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 29371845, "num_examples": 9709, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15000855, "num_examples": 4768, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20274993, "num_examples": 6604, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 183261608, "size_in_bytes": 2518585002}, "nl_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 153783851, "num_examples": 36983, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22260690, "num_examples": 5229, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 43342373, "num_examples": 10555, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22175404, "num_examples": 5171, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34259057, "num_examples": 7450, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 275821375, "size_in_bytes": 2611144769}, "nl_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 130602778, "num_examples": 41988, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18575134, "num_examples": 5866, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36296549, "num_examples": 11851, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18500040, "num_examples": 5808, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27232801, "num_examples": 8746, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 231207302, "size_in_bytes": 2566530696}, "nl_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 140654490, "num_examples": 51756, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20385605, "num_examples": 7287, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39360755, "num_examples": 14563, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20333247, "num_examples": 7229, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28939627, "num_examples": 10874, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 249673724, "size_in_bytes": 2584997118}, "nl_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 129371176, "num_examples": 41391, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18440373, "num_examples": 5774, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 36290986, "num_examples": 11691, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18363807, "num_examples": 5716, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27225846, "num_examples": 8586, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 229692188, "size_in_bytes": 2565015582}, "nl_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 196188307, "num_examples": 46283, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28400579, "num_examples": 6443, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54501140, "num_examples": 12999, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28334717, "num_examples": 6385, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45456928, "num_examples": 9894, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 352881671, "size_in_bytes": 2688205065}, "nl_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 322869564, "num_examples": 101608, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46519904, "num_examples": 14341, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 90310307, "num_examples": 28514, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35884226, "num_examples": 11090, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37816296, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 533400297, "size_in_bytes": 2868723691}, "nl_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81435686, "num_examples": 27897, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11758236, "num_examples": 3958, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22872835, "num_examples": 7959, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11652614, "num_examples": 3900, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13747983, "num_examples": 4854, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 141467354, "size_in_bytes": 2476790748}, "nl_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 178284435, "num_examples": 53401, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26010043, "num_examples": 7486, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49769059, "num_examples": 14952, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25960869, "num_examples": 7428, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37753210, "num_examples": 11058, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 317777616, "size_in_bytes": 2653101010}, "nl_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 247638902, "num_examples": 83400, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34809980, "num_examples": 11498, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67701111, "num_examples": 22978, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29281189, "num_examples": 9590, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35688498, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 415119680, "size_in_bytes": 2750443074}, "nl_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 355209659, "num_examples": 85506, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 50222904, "num_examples": 11786, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 97134573, "num_examples": 23593, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 40133479, "num_examples": 9744, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47670826, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 590371441, "size_in_bytes": 2925694835}, "nl_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 291531502, "num_examples": 94990, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41659764, "num_examples": 13299, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81223879, "num_examples": 26593, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33339983, "num_examples": 10598, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36632383, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 484387511, "size_in_bytes": 2819710905}, "nl_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 263249614, "num_examples": 86629, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37376145, "num_examples": 12024, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 72556971, "num_examples": 24116, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 32351505, "num_examples": 10370, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36080341, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 441614576, "size_in_bytes": 2776937970}, "nl_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 259048062, "num_examples": 71503, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37040268, "num_examples": 9979, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 72508834, "num_examples": 20015, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37030982, "num_examples": 9921, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43736278, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 449364424, "size_in_bytes": 2784687818}, "nl_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 311943068, "num_examples": 110300, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44231670, "num_examples": 15406, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86624232, "num_examples": 30838, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31163094, "num_examples": 10566, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34484063, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 508446127, "size_in_bytes": 2843769521}, "nl_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 184911577, "num_examples": 64035, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27150358, "num_examples": 9174, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52456241, "num_examples": 18315, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26806800, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25838330, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 317163306, "size_in_bytes": 2652486700}, "nl_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "nl_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 413357507, "num_examples": 137284, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59698683, "num_examples": 19583, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 117025047, "num_examples": 38929, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35859217, "num_examples": 11622, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36304940, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 662245394, "size_in_bytes": 2997568788}, "tr_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 27098573, "num_examples": 9234, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 3894999, "num_examples": 1290, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 7353546, "num_examples": 2550, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 3915639, "num_examples": 1290, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 7394346, "num_examples": 2550, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 49657103, "size_in_bytes": 2384980497}, "tr_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 73498845, "num_examples": 13818, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10404433, "num_examples": 1945, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20388674, "num_examples": 3954, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10435553, "num_examples": 1945, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20451938, "num_examples": 3954, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 135179443, "size_in_bytes": 2470502837}, "tr_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 50083569, "num_examples": 15861, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 7032401, "num_examples": 2248, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13702927, "num_examples": 4500, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 7068369, "num_examples": 2248, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13773759, "num_examples": 4500, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 91661025, "size_in_bytes": 2426984419}, "tr_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 45248381, "num_examples": 18769, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6527589, "num_examples": 2665, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 12524628, "num_examples": 5312, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6570229, "num_examples": 2665, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 11207680, "num_examples": 4728, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 82078507, "size_in_bytes": 2417401901}, "tr_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 48470658, "num_examples": 15778, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6912527, "num_examples": 2242, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 13549391, "num_examples": 4520, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6948399, "num_examples": 2242, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13621711, "num_examples": 4520, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 89502686, "size_in_bytes": 2424826080}, "tr_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 87989691, "num_examples": 17314, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 12878522, "num_examples": 2463, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24653361, "num_examples": 4970, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12917930, "num_examples": 2463, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24732881, "num_examples": 4970, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 163172385, "size_in_bytes": 2498495779}, "tr_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166313237, "num_examples": 51569, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24005949, "num_examples": 7379, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47270362, "num_examples": 14737, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13427829, "num_examples": 4186, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 16317167, "num_examples": 5332, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 267334544, "size_in_bytes": 2602657938}, "tr_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81435686, "num_examples": 27897, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11758236, "num_examples": 3958, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22872835, "num_examples": 7959, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11652614, "num_examples": 3900, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13747983, "num_examples": 4854, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 141467354, "size_in_bytes": 2476790748}, "tr_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 71891835, "num_examples": 20658, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10607328, "num_examples": 2964, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20228453, "num_examples": 5879, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10654752, "num_examples": 2964, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 17319632, "num_examples": 5090, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 130702000, "size_in_bytes": 2466025394}, "tr_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 121688468, "num_examples": 41635, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17389991, "num_examples": 5952, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33845052, "num_examples": 11783, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11941414, "num_examples": 4102, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14525339, "num_examples": 5140, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 199390264, "size_in_bytes": 2534713658}, "tr_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 201183647, "num_examples": 43039, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28829975, "num_examples": 6074, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 56119219, "num_examples": 12168, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18818108, "num_examples": 4090, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21735727, "num_examples": 5206, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 326686676, "size_in_bytes": 2662010070}, "tr_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 144437016, "num_examples": 47179, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20731557, "num_examples": 6709, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40731656, "num_examples": 13458, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12475286, "num_examples": 4066, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15153084, "num_examples": 5192, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 233528599, "size_in_bytes": 2568851993}, "tr_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 120543377, "num_examples": 40006, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17354201, "num_examples": 5704, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33798244, "num_examples": 11409, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12397391, "num_examples": 4108, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15077010, "num_examples": 5226, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 199170223, "size_in_bytes": 2534493617}, "tr_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 108577285, "num_examples": 27164, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15691725, "num_examples": 3861, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30777383, "num_examples": 7718, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15753501, "num_examples": 3861, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19040885, "num_examples": 5008, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 189840779, "size_in_bytes": 2525164173}, "tr_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 170406209, "num_examples": 62715, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24314937, "num_examples": 8920, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47750129, "num_examples": 17797, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11311535, "num_examples": 4138, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13996102, "num_examples": 5284, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 267778912, "size_in_bytes": 2603102306}, "tr_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 24146776, "num_examples": 9156, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 3448897, "num_examples": 1314, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 6581961, "num_examples": 2622, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 3469921, "num_examples": 1314, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 6623913, "num_examples": 2622, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 44271468, "size_in_bytes": 2379594862}, "tr_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "tr_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 253170144, "num_examples": 84795, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36250477, "num_examples": 12147, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 71755766, "num_examples": 24130, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12460985, "num_examples": 4244, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15551425, "num_examples": 5404, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 389188797, "size_in_bytes": 2724512191}, "vi_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 88462610, "num_examples": 24638, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13009081, "num_examples": 3468, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 24918723, "num_examples": 6909, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13064569, "num_examples": 3468, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22026382, "num_examples": 6120, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 161481365, "size_in_bytes": 2496804759}, "vi_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 148025719, "num_examples": 29556, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 21731708, "num_examples": 4193, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 41452971, "num_examples": 8409, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21798796, "num_examples": 4193, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38584630, "num_examples": 7620, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 271593824, "size_in_bytes": 2606917218}, "vi_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 122373588, "num_examples": 34039, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17415440, "num_examples": 4726, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33696272, "num_examples": 9493, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17491056, "num_examples": 4726, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30844107, "num_examples": 8704, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 221820463, "size_in_bytes": 2557143857}, "vi_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 120875160, "num_examples": 39555, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17565187, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33254055, "num_examples": 10909, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 17653427, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29023774, "num_examples": 9536, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 218371603, "size_in_bytes": 2553694997}, "vi_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 126974933, "num_examples": 34898, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 18308407, "num_examples": 4890, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 35737602, "num_examples": 9881, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18386647, "num_examples": 4890, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32892813, "num_examples": 9092, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 232300402, "size_in_bytes": 2567623796}, "vi_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 194630834, "num_examples": 39460, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28510169, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54262115, "num_examples": 11087, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28598409, "num_examples": 5515, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51436622, "num_examples": 10298, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 357438149, "size_in_bytes": 2692761543}, "vi_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 278142529, "num_examples": 79631, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40387383, "num_examples": 11283, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78432015, "num_examples": 22470, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29871727, "num_examples": 8090, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43587836, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 470421490, "size_in_bytes": 2805744884}, "vi_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 178284435, "num_examples": 53401, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26010043, "num_examples": 7486, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49769059, "num_examples": 14952, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25960869, "num_examples": 7428, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37753210, "num_examples": 11058, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 317777616, "size_in_bytes": 2653101010}, "vi_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 71891835, "num_examples": 20658, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 10607328, "num_examples": 2964, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 20228453, "num_examples": 5879, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 10654752, "num_examples": 2964, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 17319632, "num_examples": 5090, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 130702000, "size_in_bytes": 2466025394}, "vi_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 212943490, "num_examples": 65165, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 30785710, "num_examples": 9186, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 59228340, "num_examples": 18234, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25388877, "num_examples": 7336, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37008958, "num_examples": 10802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 365355375, "size_in_bytes": 2700678769}, "vi_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 309096244, "num_examples": 66797, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44678886, "num_examples": 9370, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86538146, "num_examples": 18769, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34719755, "num_examples": 7386, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49257385, "num_examples": 11018, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 524290416, "size_in_bytes": 2859613810}, "vi_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 246221190, "num_examples": 73037, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35614204, "num_examples": 10311, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69208578, "num_examples": 20619, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27415565, "num_examples": 7668, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40741697, "num_examples": 11564, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 419201234, "size_in_bytes": 2754524628}, "vi_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 222640494, "num_examples": 66232, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32266832, "num_examples": 9316, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62284448, "num_examples": 18650, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27367814, "num_examples": 7720, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40676185, "num_examples": 11678, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 385235773, "size_in_bytes": 2720559167}, "vi_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 227644180, "num_examples": 55074, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33139736, "num_examples": 7755, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64340372, "num_examples": 15509, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33263816, "num_examples": 7755, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49696977, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 408085081, "size_in_bytes": 2743408475}, "vi_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 270783437, "num_examples": 89047, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 38976633, "num_examples": 12590, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 76207835, "num_examples": 25136, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26031951, "num_examples": 7808, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39568347, "num_examples": 11834, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 451568203, "size_in_bytes": 2786891597}, "vi_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 139143630, "num_examples": 39786, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20620040, "num_examples": 5676, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39773153, "num_examples": 11367, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20710856, "num_examples": 5676, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31669407, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 251917086, "size_in_bytes": 2587240480}, "vi_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "vi_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 366496821, "num_examples": 113929, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 53156359, "num_examples": 16303, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 104148958, "num_examples": 32353, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29433363, "num_examples": 8400, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42018643, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 595254144, "size_in_bytes": 2930577538}, "it_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 139415715, "num_examples": 46519, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20028591, "num_examples": 6592, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39070072, "num_examples": 13123, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14590254, "num_examples": 4742, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19771799, "num_examples": 6480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 232876431, "size_in_bytes": 2568199825}, "it_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 192953917, "num_examples": 49985, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27761700, "num_examples": 7103, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53523513, "num_examples": 14077, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22331539, "num_examples": 5253, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34240504, "num_examples": 7434, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 330811173, "size_in_bytes": 2666134567}, "it_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 167429228, "num_examples": 54520, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23647408, "num_examples": 7676, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46393975, "num_examples": 15377, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18226415, "num_examples": 5826, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27130598, "num_examples": 8734, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 282827624, "size_in_bytes": 2618151018}, "it_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 177916880, "num_examples": 64406, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 25411901, "num_examples": 9101, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 49344464, "num_examples": 18055, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20013708, "num_examples": 7251, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28723163, "num_examples": 10828, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 301410116, "size_in_bytes": 2636733510}, "it_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166337969, "num_examples": 54061, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23603565, "num_examples": 7606, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46541821, "num_examples": 15269, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18181452, "num_examples": 5756, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27277884, "num_examples": 8626, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 281942691, "size_in_bytes": 2617266085}, "it_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 229125770, "num_examples": 58235, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33012428, "num_examples": 8183, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 63609914, "num_examples": 16337, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27599547, "num_examples": 6333, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44363065, "num_examples": 9694, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 397710724, "size_in_bytes": 2733034118}, "it_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 408425072, "num_examples": 129572, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55208131, "num_examples": 17277, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 108498129, "num_examples": 34496, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38427450, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37873247, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 648432029, "size_in_bytes": 2983755423}, "it_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 247638902, "num_examples": 83400, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34809980, "num_examples": 11498, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67701111, "num_examples": 22978, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29281189, "num_examples": 9590, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35688498, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 415119680, "size_in_bytes": 2750443074}, "it_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 121688468, "num_examples": 41635, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17389991, "num_examples": 5952, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33845052, "num_examples": 11783, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11941414, "num_examples": 4102, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 14525339, "num_examples": 5140, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 199390264, "size_in_bytes": 2534713658}, "it_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 212943490, "num_examples": 65165, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 30785710, "num_examples": 9186, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 59228340, "num_examples": 18234, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25388877, "num_examples": 7336, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37008958, "num_examples": 10802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 365355375, "size_in_bytes": 2700678769}, "it_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 466063907, "num_examples": 117200, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 62368027, "num_examples": 15446, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 120527290, "num_examples": 30583, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46962303, "num_examples": 11612, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47517662, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 743439189, "size_in_bytes": 3078762583}, "it_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 375454934, "num_examples": 123262, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 50391870, "num_examples": 16397, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 98678549, "num_examples": 32609, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36746798, "num_examples": 11904, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36576135, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 597848286, "size_in_bytes": 2933171680}, "it_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 326270925, "num_examples": 107461, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44585547, "num_examples": 14480, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86601736, "num_examples": 28804, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34225344, "num_examples": 11034, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36584411, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 528267963, "size_in_bytes": 2863591357}, "it_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 283078038, "num_examples": 80071, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40067356, "num_examples": 11109, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78572021, "num_examples": 22241, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34701291, "num_examples": 9259, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44205612, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 480624318, "size_in_bytes": 2815947712}, "it_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 412122842, "num_examples": 145546, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56845599, "num_examples": 20004, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 111650085, "num_examples": 39914, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34601229, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33981021, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 649200776, "size_in_bytes": 2984524170}, "it_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 291344418, "num_examples": 102255, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41526629, "num_examples": 14550, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81244342, "num_examples": 28929, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25705667, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25298358, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 465119414, "size_in_bytes": 2800442808}, "it_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "it_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 521360992, "num_examples": 174428, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 73870968, "num_examples": 24671, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 145529087, "num_examples": 49049, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36056517, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35512219, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 812329783, "size_in_bytes": 3147653177}, "ru_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 222796210, "num_examples": 47777, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32376186, "num_examples": 6780, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62643370, "num_examples": 13486, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22375615, "num_examples": 4796, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28280966, "num_examples": 6524, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 368472347, "size_in_bytes": 2703795741}, "ru_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 279329223, "num_examples": 51611, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40283523, "num_examples": 7303, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 77675220, "num_examples": 14536, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 30291320, "num_examples": 5319, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43329616, "num_examples": 7574, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 470908902, "size_in_bytes": 2806232296}, "ru_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 256536185, "num_examples": 56270, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36624583, "num_examples": 7914, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 71046411, "num_examples": 15814, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26642156, "num_examples": 5930, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36720087, "num_examples": 8852, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 427569422, "size_in_bytes": 2762892816}, "ru_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 271938701, "num_examples": 66166, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 39039165, "num_examples": 9317, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 75905428, "num_examples": 18614, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29079186, "num_examples": 7333, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40223132, "num_examples": 11068, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 456185612, "size_in_bytes": 2791509006}, "ru_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 254388387, "num_examples": 55519, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36467209, "num_examples": 7814, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 71219647, "num_examples": 15682, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26483182, "num_examples": 5830, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36892379, "num_examples": 8720, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 425450804, "size_in_bytes": 2760774198}, "ru_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 319843264, "num_examples": 59843, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46169619, "num_examples": 8373, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 89184436, "num_examples": 16808, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36194536, "num_examples": 6389, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 54875184, "num_examples": 9846, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 546267039, "size_in_bytes": 2881590433}, "ru_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 544791012, "num_examples": 132550, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 72902906, "num_examples": 17431, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 141797425, "num_examples": 34639, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 51228388, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 50153085, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 860872816, "size_in_bytes": 3196196210}, "ru_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 355209659, "num_examples": 85506, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 50222904, "num_examples": 11786, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 97134573, "num_examples": 23593, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 40133479, "num_examples": 9744, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47670826, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 590371441, "size_in_bytes": 2925694835}, "ru_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 201183647, "num_examples": 43039, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28829975, "num_examples": 6074, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 56119219, "num_examples": 12168, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18818108, "num_examples": 4090, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21735727, "num_examples": 5206, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 326686676, "size_in_bytes": 2662010070}, "ru_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 309096244, "num_examples": 66797, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44678886, "num_examples": 9370, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86538146, "num_examples": 18769, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34719755, "num_examples": 7386, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49257385, "num_examples": 11018, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 524290416, "size_in_bytes": 2859613810}, "ru_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 466063907, "num_examples": 117200, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 62368027, "num_examples": 15446, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 120527290, "num_examples": 30583, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46962303, "num_examples": 11612, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47517662, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 743439189, "size_in_bytes": 3078762583}, "ru_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 508123956, "num_examples": 126078, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 67376079, "num_examples": 16425, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 130501807, "num_examples": 32556, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 49166213, "num_examples": 11798, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49034200, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 804202255, "size_in_bytes": 3139525649}, "ru_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 449251907, "num_examples": 109927, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 61599020, "num_examples": 14804, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 119178429, "num_examples": 29491, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46678759, "num_examples": 11224, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 48633663, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 725341778, "size_in_bytes": 3060665172}, "ru_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 387730029, "num_examples": 81807, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55246840, "num_examples": 11387, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 108002867, "num_examples": 22722, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 45319981, "num_examples": 9403, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 56802711, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 653102428, "size_in_bytes": 2988425822}, "ru_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 549199449, "num_examples": 147170, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 74700030, "num_examples": 19776, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 146797401, "num_examples": 39675, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 47703025, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46899896, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 865299801, "size_in_bytes": 3200623195}, "ru_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 447429810, "num_examples": 105939, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64206498, "num_examples": 14952, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 125223057, "num_examples": 29886, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38729780, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37713168, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 713302313, "size_in_bytes": 3048625707}, "ru_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ru_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 667427997, "num_examples": 176524, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 94191485, "num_examples": 24639, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 184829601, "num_examples": 49162, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 49103023, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 48209624, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 1043761730, "size_in_bytes": 3379085124}, "de_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 164086715, "num_examples": 52293, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23764268, "num_examples": 7431, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46438098, "num_examples": 14834, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15519549, "num_examples": 4788, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20881542, "num_examples": 6568, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 270690172, "size_in_bytes": 2606013566}, "de_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221028649, "num_examples": 56409, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32022615, "num_examples": 8040, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62270935, "num_examples": 16080, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23787640, "num_examples": 5397, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36734315, "num_examples": 7814, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 375844154, "size_in_bytes": 2711167548}, "de_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 196039909, "num_examples": 61638, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 28068370, "num_examples": 8725, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 55190893, "num_examples": 17542, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19844355, "num_examples": 6082, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29676497, "num_examples": 9276, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 328820024, "size_in_bytes": 2664143418}, "de_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 207428120, "num_examples": 71914, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 29904137, "num_examples": 10184, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 58657522, "num_examples": 20396, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21703466, "num_examples": 7541, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31788018, "num_examples": 11546, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 349481263, "size_in_bytes": 2684804657}, "de_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 193559597, "num_examples": 60607, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27659067, "num_examples": 8557, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54581778, "num_examples": 17236, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19432364, "num_examples": 5914, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29063654, "num_examples": 8970, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 324296460, "size_in_bytes": 2659619854}, "de_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 260545427, "num_examples": 65317, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37782101, "num_examples": 9220, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 73264764, "num_examples": 18524, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29566006, "num_examples": 6577, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47767248, "num_examples": 10258, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 448925546, "size_in_bytes": 2784248940}, "de_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 478710842, "num_examples": 148610, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64658873, "num_examples": 19754, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 127179527, "num_examples": 39359, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39564815, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38692244, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 748806301, "size_in_bytes": 3084129695}, "de_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 291531502, "num_examples": 94990, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41659764, "num_examples": 13299, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81223879, "num_examples": 26593, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33339983, "num_examples": 10598, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36632383, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 484387511, "size_in_bytes": 2819710905}, "de_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 144437016, "num_examples": 47179, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20731557, "num_examples": 6709, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 40731656, "num_examples": 13458, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12475286, "num_examples": 4066, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15153084, "num_examples": 5192, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 233528599, "size_in_bytes": 2568851993}, "de_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 246221190, "num_examples": 73037, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35614204, "num_examples": 10311, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69208578, "num_examples": 20619, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27415565, "num_examples": 7668, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40741697, "num_examples": 11564, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 419201234, "size_in_bytes": 2754524628}, "de_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 375454934, "num_examples": 123262, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 50391870, "num_examples": 16397, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 98678549, "num_examples": 32609, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36746798, "num_examples": 11904, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36576135, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 597848286, "size_in_bytes": 2933171680}, "de_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 508123956, "num_examples": 126078, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 67376079, "num_examples": 16425, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 130501807, "num_examples": 32556, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 49166213, "num_examples": 11798, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49034200, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 804202255, "size_in_bytes": 3139525649}, "de_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 377445210, "num_examples": 120829, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 52558148, "num_examples": 16591, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 102649431, "num_examples": 33139, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38222697, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37463495, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 608338981, "size_in_bytes": 2943662375}, "de_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 326604493, "num_examples": 90351, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46761156, "num_examples": 12704, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91791176, "num_examples": 25380, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38600805, "num_examples": 10061, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45391392, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 549149022, "size_in_bytes": 2884472416}, "de_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 461493612, "num_examples": 158360, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 62415266, "num_examples": 21233, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 122636824, "num_examples": 42327, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35863347, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35061714, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 717470763, "size_in_bytes": 3052794157}, "de_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 352643325, "num_examples": 118515, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 50842582, "num_examples": 16929, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 100089912, "num_examples": 33798, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27151032, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26769006, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 557495857, "size_in_bytes": 2892819251}, "de_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "de_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 581398440, "num_examples": 189844, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 82047662, "num_examples": 26618, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 162086569, "num_examples": 52978, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37311477, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37158518, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 900002666, "size_in_bytes": 3235326060}, "id_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 140142281, "num_examples": 45146, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20241248, "num_examples": 6382, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39456250, "num_examples": 12801, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15295286, "num_examples": 4786, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 20757288, "num_examples": 6618, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 235892353, "size_in_bytes": 2571215747}, "id_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 195370301, "num_examples": 48962, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27987737, "num_examples": 6903, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 54356009, "num_examples": 13905, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23050111, "num_examples": 5307, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35674711, "num_examples": 7722, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 336438869, "size_in_bytes": 2671762263}, "id_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 169716742, "num_examples": 53785, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23942457, "num_examples": 7514, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47145159, "num_examples": 15213, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19014607, "num_examples": 5918, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28483621, "num_examples": 9030, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 288302586, "size_in_bytes": 2623625980}, "id_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 180732752, "num_examples": 64011, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 26075830, "num_examples": 9053, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 50607034, "num_examples": 18051, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 21172604, "num_examples": 7457, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 30590132, "num_examples": 11284, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 309178352, "size_in_bytes": 2644501746}, "id_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 170279570, "num_examples": 53748, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24544740, "num_examples": 7592, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47749544, "num_examples": 15273, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19618138, "num_examples": 5996, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29090134, "num_examples": 9090, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 291282126, "size_in_bytes": 2626605520}, "id_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 236710901, "num_examples": 58252, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34194921, "num_examples": 8175, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65735448, "num_examples": 16439, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29277647, "num_examples": 6579, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 47094694, "num_examples": 10256, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 413013611, "size_in_bytes": 2748337005}, "id_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 419393003, "num_examples": 129987, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59255562, "num_examples": 18095, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 116049787, "num_examples": 36136, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 39446774, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38401219, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 672546345, "size_in_bytes": 3007869739}, "id_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 263249614, "num_examples": 86629, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37376145, "num_examples": 12024, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 72556971, "num_examples": 24116, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 32351505, "num_examples": 10370, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36080341, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 441614576, "size_in_bytes": 2776937970}, "id_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 120543377, "num_examples": 40006, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 17354201, "num_examples": 5704, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 33798244, "num_examples": 11409, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12397391, "num_examples": 4108, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15077010, "num_examples": 5226, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 199170223, "size_in_bytes": 2534493617}, "id_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 222640494, "num_examples": 66232, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 32266832, "num_examples": 9316, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62284448, "num_examples": 18650, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27367814, "num_examples": 7720, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 40676185, "num_examples": 11678, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 385235773, "size_in_bytes": 2720559167}, "id_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 326270925, "num_examples": 107461, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44585547, "num_examples": 14480, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86601736, "num_examples": 28804, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34225344, "num_examples": 11034, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36584411, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 528267963, "size_in_bytes": 2863591357}, "id_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 449251907, "num_examples": 109927, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 61599020, "num_examples": 14804, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 119178429, "num_examples": 29491, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 46678759, "num_examples": 11224, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 48633663, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 725341778, "size_in_bytes": 3060665172}, "id_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 377445210, "num_examples": 120829, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 52558148, "num_examples": 16591, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 102649431, "num_examples": 33139, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38222697, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37463495, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 608338981, "size_in_bytes": 2943662375}, "id_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 301212454, "num_examples": 82970, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 43181406, "num_examples": 11615, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 84344406, "num_examples": 23329, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38319172, "num_examples": 10019, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44804580, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 511862018, "size_in_bytes": 2847185412}, "id_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 397744429, "num_examples": 136931, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55512600, "num_examples": 18914, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 108612101, "num_examples": 37750, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35961493, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35361450, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 633192073, "size_in_bytes": 2968515467}, "id_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 287661021, "num_examples": 96684, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41606399, "num_examples": 13788, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81161727, "num_examples": 27549, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27204723, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26865286, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 464499156, "size_in_bytes": 2799822550}, "id_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "id_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 514604836, "num_examples": 168159, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 73631130, "num_examples": 23839, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 144464490, "num_examples": 47431, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37499169, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37146873, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 807346498, "size_in_bytes": 3142669892}, "ar_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 131413705, "num_examples": 32728, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 19265725, "num_examples": 4649, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 37437193, "num_examples": 9266, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 19340109, "num_examples": 4649, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25725463, "num_examples": 6556, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 233182195, "size_in_bytes": 2568505589}, "ar_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 187631967, "num_examples": 36518, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27287921, "num_examples": 5162, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52992331, "num_examples": 10366, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27370513, "num_examples": 5162, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 41298201, "num_examples": 7656, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 336580933, "size_in_bytes": 2671904327}, "ar_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 164263852, "num_examples": 41471, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 23242384, "num_examples": 5771, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 45970025, "num_examples": 11638, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 23334720, "num_examples": 5771, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34295079, "num_examples": 8928, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 291106060, "size_in_bytes": 2626429454}, "ar_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 172246002, "num_examples": 49583, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24970206, "num_examples": 7004, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 48502033, "num_examples": 13938, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25082270, "num_examples": 7004, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35463115, "num_examples": 10644, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 306263626, "size_in_bytes": 2641587020}, "ar_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 166772243, "num_examples": 41564, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24045339, "num_examples": 5859, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 46975608, "num_examples": 11680, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24139083, "num_examples": 5859, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35302502, "num_examples": 8970, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 297234775, "size_in_bytes": 2632558169}, "ar_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 238218279, "num_examples": 46708, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 34684531, "num_examples": 6538, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 67280813, "num_examples": 13190, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34789139, "num_examples": 6538, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 55631867, "num_examples": 10480, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 430604629, "size_in_bytes": 2765928023}, "ar_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 363338130, "num_examples": 97839, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 52510566, "num_examples": 13880, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 102637261, "num_examples": 27553, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42036462, "num_examples": 10687, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46790994, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 607313413, "size_in_bytes": 2942636807}, "ar_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 259048062, "num_examples": 71503, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 37040268, "num_examples": 9979, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 72508834, "num_examples": 20015, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37030982, "num_examples": 9921, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43736278, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 449364424, "size_in_bytes": 2784687818}, "ar_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 108577285, "num_examples": 27164, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 15691725, "num_examples": 3861, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 30777383, "num_examples": 7718, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15753501, "num_examples": 3861, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19040885, "num_examples": 5008, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 189840779, "size_in_bytes": 2525164173}, "ar_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 227644180, "num_examples": 55074, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33139736, "num_examples": 7755, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64340372, "num_examples": 15509, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33263816, "num_examples": 7755, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 49696977, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 408085081, "size_in_bytes": 2743408475}, "ar_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 283078038, "num_examples": 80071, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 40067356, "num_examples": 11109, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78572021, "num_examples": 22241, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34701291, "num_examples": 9259, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44205612, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 480624318, "size_in_bytes": 2815947712}, "ar_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 387730029, "num_examples": 81807, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55246840, "num_examples": 11387, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 108002867, "num_examples": 22722, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 45319981, "num_examples": 9403, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 56802711, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 653102428, "size_in_bytes": 2988425822}, "ar_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 326604493, "num_examples": 90351, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46761156, "num_examples": 12704, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91791176, "num_examples": 25380, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38600805, "num_examples": 10061, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45391392, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 549149022, "size_in_bytes": 2884472416}, "ar_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 301212454, "num_examples": 82970, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 43181406, "num_examples": 11615, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 84344406, "num_examples": 23329, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38319172, "num_examples": 10019, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 44804580, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 511862018, "size_in_bytes": 2847185412}, "ar_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 349815434, "num_examples": 106423, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 49774892, "num_examples": 14955, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 98219457, "num_examples": 29875, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36868050, "num_examples": 10173, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42941673, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 577619506, "size_in_bytes": 2912942900}, "ar_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 228851585, "num_examples": 59976, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33475038, "num_examples": 8577, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65673225, "num_examples": 17130, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33612270, "num_examples": 8577, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34545442, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 396157560, "size_in_bytes": 2731480954}, "ar_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "ar_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 455185296, "num_examples": 133295, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 66013945, "num_examples": 19096, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129585745, "num_examples": 37832, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42335637, "num_examples": 11193, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45021748, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 738142371, "size_in_bytes": 3073465765}, "pt_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 188786420, "num_examples": 67687, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27128353, "num_examples": 9624, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 53313973, "num_examples": 19197, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 14136215, "num_examples": 4842, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 19582346, "num_examples": 6684, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 302947307, "size_in_bytes": 2638270701}, "pt_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 247952847, "num_examples": 72147, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 35719383, "num_examples": 10293, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 69647118, "num_examples": 20479, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22737949, "num_examples": 5511, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35936003, "num_examples": 7966, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 411993300, "size_in_bytes": 2747316694}, "pt_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 221794282, "num_examples": 77592, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31524351, "num_examples": 10986, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 62126672, "num_examples": 21977, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18554005, "num_examples": 6204, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28438357, "num_examples": 9464, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 362437667, "size_in_bytes": 2697761061}, "pt_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 230787450, "num_examples": 87412, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33050829, "num_examples": 12369, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 64700437, "num_examples": 24669, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20102611, "num_examples": 7587, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 29654422, "num_examples": 11572, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 378295749, "size_in_bytes": 2713619143}, "pt_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 219249985, "num_examples": 76567, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 31333782, "num_examples": 10850, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 61684803, "num_examples": 21707, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 18361260, "num_examples": 6068, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 27993336, "num_examples": 9194, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 358623166, "size_in_bytes": 2693946560}, "pt_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 286289876, "num_examples": 81039, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41347017, "num_examples": 11437, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 80409979, "num_examples": 22905, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28383887, "num_examples": 6655, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46737680, "num_examples": 10392, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 483168439, "size_in_bytes": 2818491833}, "pt_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 503678107, "num_examples": 167390, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 66995531, "num_examples": 21979, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 131128053, "num_examples": 43578, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37166280, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36642991, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 775610962, "size_in_bytes": 3110934356}, "pt_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 311943068, "num_examples": 110300, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44231670, "num_examples": 15406, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 86624232, "num_examples": 30838, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31163094, "num_examples": 10566, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34484063, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 508446127, "size_in_bytes": 2843769521}, "pt_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 170406209, "num_examples": 62715, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 24314937, "num_examples": 8920, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 47750129, "num_examples": 17797, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11311535, "num_examples": 4138, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 13996102, "num_examples": 5284, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 267778912, "size_in_bytes": 2603102306}, "pt_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 270783437, "num_examples": 89047, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 38976633, "num_examples": 12590, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 76207835, "num_examples": 25136, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26031951, "num_examples": 7808, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39568347, "num_examples": 11834, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 451568203, "size_in_bytes": 2786891597}, "pt_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 412122842, "num_examples": 145546, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 56845599, "num_examples": 20004, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 111650085, "num_examples": 39914, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34601229, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 33981021, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 649200776, "size_in_bytes": 2984524170}, "pt_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 549199449, "num_examples": 147170, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 74700030, "num_examples": 19776, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 146797401, "num_examples": 39675, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 47703025, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 46899896, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 865299801, "size_in_bytes": 3200623195}, "pt_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 461493612, "num_examples": 158360, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 62415266, "num_examples": 21233, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 122636824, "num_examples": 42327, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35863347, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35061714, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 717470763, "size_in_bytes": 3052794157}, "pt_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 397744429, "num_examples": 136931, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55512600, "num_examples": 18914, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 108612101, "num_examples": 37750, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35961493, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35361450, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 633192073, "size_in_bytes": 2968515467}, "pt_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 349815434, "num_examples": 106423, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 49774892, "num_examples": 14955, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 98219457, "num_examples": 29875, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36868050, "num_examples": 10173, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42941673, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 577619506, "size_in_bytes": 2912942900}, "pt_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 443141327, "num_examples": 164019, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 63289451, "num_examples": 23346, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 124735747, "num_examples": 46539, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24528320, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24122878, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 679817723, "size_in_bytes": 3015141117}, "pt_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "pt_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 677523842, "num_examples": 235172, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 95695244, "num_examples": 33073, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 176843858, "num_examples": 61369, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34633311, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34546285, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 1019242540, "size_in_bytes": 3354565934}, "en_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 44453638, "num_examples": 14616, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 6612386, "num_examples": 2082, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 12756303, "num_examples": 4164, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 6645698, "num_examples": 2082, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 12822927, "num_examples": 4164, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 83290952, "size_in_bytes": 2418614346}, "en_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 110618152, "num_examples": 20172, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 16073147, "num_examples": 2883, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 31274304, "num_examples": 5802, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 16119275, "num_examples": 2883, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31367136, "num_examples": 5802, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 205452014, "size_in_bytes": 2540775408}, "en_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81409319, "num_examples": 25971, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11560905, "num_examples": 3678, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22773109, "num_examples": 7440, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11619753, "num_examples": 3678, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22890981, "num_examples": 7440, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 150254067, "size_in_bytes": 2485577461}, "en_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 92487460, "num_examples": 37569, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 13596249, "num_examples": 5391, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 26321891, "num_examples": 10752, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 13682505, "num_examples": 5391, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22059198, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 168147303, "size_in_bytes": 2503470697}, "en_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 79823798, "num_examples": 25110, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 11488843, "num_examples": 3564, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 22564617, "num_examples": 7188, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 11545867, "num_examples": 3564, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 22679625, "num_examples": 7188, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 148102750, "size_in_bytes": 2483426144}, "en_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 152727050, "num_examples": 30174, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 22743105, "num_examples": 4311, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 43247373, "num_examples": 8628, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22812081, "num_examples": 4311, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 43385421, "num_examples": 8628, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 284915030, "size_in_bytes": 2620238424}, "en_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 404347028, "num_examples": 130269, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 58305760, "num_examples": 18579, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 115530811, "num_examples": 37215, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 28395953, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 28017229, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 634596781, "size_in_bytes": 2969920175}, "en_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 184911577, "num_examples": 64035, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 27150358, "num_examples": 9174, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 52456241, "num_examples": 18315, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26806800, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25838330, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 317163306, "size_in_bytes": 2652486700}, "en_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 24146776, "num_examples": 9156, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 3448897, "num_examples": 1314, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 6581961, "num_examples": 2622, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 3469921, "num_examples": 1314, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 6623913, "num_examples": 2622, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 44271468, "size_in_bytes": 2379594862}, "en_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 139143630, "num_examples": 39786, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 20620040, "num_examples": 5676, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 39773153, "num_examples": 11367, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20710856, "num_examples": 5676, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31669407, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 251917086, "size_in_bytes": 2587240480}, "en_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 291344418, "num_examples": 102255, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41526629, "num_examples": 14550, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81244342, "num_examples": 28929, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 25705667, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 25298358, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 465119414, "size_in_bytes": 2800442808}, "en_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 447429810, "num_examples": 105939, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 64206498, "num_examples": 14952, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 125223057, "num_examples": 29886, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38729780, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37713168, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 713302313, "size_in_bytes": 3048625707}, "en_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 352643325, "num_examples": 118515, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 50842582, "num_examples": 16929, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 100089912, "num_examples": 33798, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27151032, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26769006, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 557495857, "size_in_bytes": 2892819251}, "en_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 287661021, "num_examples": 96684, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 41606399, "num_examples": 13788, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 81161727, "num_examples": 27549, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 27204723, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26865286, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 464499156, "size_in_bytes": 2799822550}, "en_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 228851585, "num_examples": 59976, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 33475038, "num_examples": 8577, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 65673225, "num_examples": 17130, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 33612270, "num_examples": 8577, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34545442, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 396157560, "size_in_bytes": 2731480954}, "en_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 443141327, "num_examples": 164019, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 63289451, "num_examples": 23346, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 124735747, "num_examples": 46539, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24528320, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 24122878, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 679817723, "size_in_bytes": 3015141117}, "en_es": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "en_es", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 662074792, "num_examples": 228885, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 94598261, "num_examples": 32709, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 188640095, "num_examples": 65178, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26008038, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26622402, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 997943588, "size_in_bytes": 3333266982}, "es_cs": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_cs", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 274283144, "num_examples": 90407, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 39462727, "num_examples": 12911, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 78171714, "num_examples": 25720, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 15685459, "num_examples": 5008, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 21992813, "num_examples": 6994, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 429595857, "size_in_bytes": 2764919251}, "es_hi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_hi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 336530602, "num_examples": 95119, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 48593955, "num_examples": 13628, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 95243183, "num_examples": 27064, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 24828159, "num_examples": 5725, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 39085786, "num_examples": 8338, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 544281685, "size_in_bytes": 2879605079}, "es_ja": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ja", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 309354143, "num_examples": 100740, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44193203, "num_examples": 14375, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 87558590, "num_examples": 28660, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20439359, "num_examples": 6472, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31425561, "num_examples": 9934, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 492970856, "size_in_bytes": 2828294250}, "es_zh": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_zh", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 321867278, "num_examples": 112030, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 46353595, "num_examples": 16036, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 91572010, "num_examples": 31906, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 22626327, "num_examples": 8133, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 32425726, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 514844936, "size_in_bytes": 2850168330}, "es_ko": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ko", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 308342677, "num_examples": 100095, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 44330902, "num_examples": 14315, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 87641207, "num_examples": 28502, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 20576098, "num_examples": 6412, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 31506818, "num_examples": 9776, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 492397702, "size_in_bytes": 2827721096}, "es_th": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_th", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 379460557, "num_examples": 104975, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 55164438, "num_examples": 15010, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 107476969, "num_examples": 29824, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 31420754, "num_examples": 7107, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 51363732, "num_examples": 11098, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 624886450, "size_in_bytes": 2960209844}, "es_fr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_fr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 632129442, "num_examples": 200858, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 89191353, "num_examples": 28040, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 175918668, "num_examples": 55753, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 38602577, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 38440897, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 974282937, "size_in_bytes": 3309606331}, "es_nl": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_nl", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 413357507, "num_examples": 137284, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 59698683, "num_examples": 19583, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 117025047, "num_examples": 38929, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 35859217, "num_examples": 11622, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 36304940, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 662245394, "size_in_bytes": 2997568788}, "es_tr": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_tr", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 253170144, "num_examples": 84795, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 36250477, "num_examples": 12147, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 71755766, "num_examples": 24130, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 12460985, "num_examples": 4244, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 15551425, "num_examples": 5404, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 389188797, "size_in_bytes": 2724512191}, "es_vi": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_vi", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 366496821, "num_examples": 113929, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 53156359, "num_examples": 16303, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 104148958, "num_examples": 32353, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 29433363, "num_examples": 8400, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 42018643, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 595254144, "size_in_bytes": 2930577538}, "es_it": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_it", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 521360992, "num_examples": 174428, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 73870968, "num_examples": 24671, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 145529087, "num_examples": 49049, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 36056517, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 35512219, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 812329783, "size_in_bytes": 3147653177}, "es_ru": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ru", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 667427997, "num_examples": 176524, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 94191485, "num_examples": 24639, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 184829601, "num_examples": 49162, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 49103023, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 48209624, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 1043761730, "size_in_bytes": 3379085124}, "es_de": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_de", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 581398440, "num_examples": 189844, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 82047662, "num_examples": 26618, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 162086569, "num_examples": 52978, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37311477, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37158518, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 900002666, "size_in_bytes": 3235326060}, "es_id": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_id", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 514604836, "num_examples": 168159, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 73631130, "num_examples": 23839, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 144464490, "num_examples": 47431, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 37499169, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 37146873, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 807346498, "size_in_bytes": 3142669892}, "es_ar": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_ar", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 455185296, "num_examples": 133295, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 66013945, "num_examples": 19096, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 129585745, "num_examples": 37832, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 42335637, "num_examples": 11193, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 45021748, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 738142371, "size_in_bytes": 3073465765}, "es_pt": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_pt", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 677523842, "num_examples": 235172, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 95695244, "num_examples": 33073, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 176843858, "num_examples": 61369, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 34633311, "num_examples": 12000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 34546285, "num_examples": 12000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 1019242540, "size_in_bytes": 3354565934}, "es_en": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "es_en", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 662074792, "num_examples": 228885, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 94598261, "num_examples": 32709, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 188640095, "num_examples": 65178, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 26008038, "num_examples": 9000, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 26622402, "num_examples": 9000, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 997943588, "size_in_bytes": 3333266982}, "multilingual": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "multilingual", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1692034240, "num_examples": 519894, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 243053420, "num_examples": 73841, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 479801485, "num_examples": 148323, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 139809604, "num_examples": 39492, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 171558547, "num_examples": 47948, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 2726257296, "size_in_bytes": 5061580690}, "crosslingual": {"description": "WikiLingua is a large-scale multilingual dataset for the evaluation of\ncrosslingual abstractive summarization systems. The dataset includes ~770k\narticle and summary pairs in 18 languages from WikiHow. The gold-standard\narticle-summary alignments across languages was done by aligning the images\nthat are used to describe each how-to step in an article.\n", "citation": "@article{ladhak-wiki-2020,\n title = {WikiLingua: A New Benchmark Dataset for Multilingual Abstractive Summarization},\n authors = {Faisal Ladhak, Esin Durmus, Claire Cardie and Kathleen McKeown},\n journal = {arXiv preprint arXiv:2010.03093},\n year = {2020},\n url = {https://arxiv.org/abs/2010.03093}\n}\n", "homepage": "https://github.com/esdurmus/Wikilingua", "license": "CC BY-NC-SA 3.0", "features": {"gem_id": {"dtype": "string", "id": null, "_type": "Value"}, "gem_parent_id": {"dtype": "string", "id": null, "_type": "Value"}, "source_language": {"dtype": "string", "id": null, "_type": "Value"}, "target_language": {"dtype": "string", "id": null, "_type": "Value"}, "source": {"dtype": "string", "id": null, "_type": "Value"}, "target": {"dtype": "string", "id": null, "_type": "Value"}, "references": [{"dtype": "string", "id": null, "_type": "Value"}]}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "wiki_lingua", "config_name": "crosslingual", "version": {"version_str": "2.0.0", "description": null, "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 12829211854, "num_examples": 3745122, "dataset_name": "wiki_lingua"}, "validation": {"name": "validation", "num_bytes": 1771158322, "num_examples": 505610, "dataset_name": "wiki_lingua"}, "test": {"name": "test", "num_bytes": 3419038963, "num_examples": 1004204, "dataset_name": "wiki_lingua"}, "sampled_validation": {"name": "sampled_validation", "num_bytes": 1520650050, "num_examples": 421384, "dataset_name": "wiki_lingua"}, "sampled_test": {"name": "sampled_test", "num_bytes": 2226407222, "num_examples": 623510, "dataset_name": "wiki_lingua"}}, "download_checksums": {"wikilingua_cleaned.tar.gz": {"num_bytes": 2335323394, "checksum": "95663019cebc422c9b017f6b670a67a272eef259ce69d63677bf4606b692ed83"}}, "download_size": 2335323394, "post_processing_size": null, "dataset_size": 21766466411, "size_in_bytes": 24101789805}} \ No newline at end of file