Update files from the datasets library (from 1.1.3)

Release notes: https://github.com/huggingface/datasets/releases/tag/1.1.3

dataset_infos.json

@@ -1 +1 @@
-{"plain_text": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"languages": ["ar", "bg", "de", "el", "en", "es", "fr", "hi", "ru", "sw", "th", "tr", "ur", "vi", "zh"], "id": null, "_type": "Translation"}, "hypothesis": {"languages": ["ar", "bg", "de", "el", "en", "es", "fr", "hi", "ru", "sw", "th", "tr", "ur", "vi", "zh"], "num_languages": 15, "id": null, "_type": "TranslationVariableLanguages"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "plain_text", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"test": {"name": "test", "num_bytes": 19387508, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 9566255, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://cims.nyu.edu/~sbowman/xnli/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 17865352, "post_processing_size": null, "dataset_size": 28953763, "size_in_bytes": 46819115}}
+{"ar": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "ar", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 107399934, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 1294561, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 633009, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 109327504, "size_in_bytes": 593291216}, "bg": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "bg", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 125973545, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 1573042, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 774069, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 128320656, "size_in_bytes": 612284368}, "de": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "de", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 84684460, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 996496, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 494612, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 86175568, "size_in_bytes": 570139280}, "el": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "el", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 139753678, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 1704793, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 841234, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 142299705, "size_in_bytes": 626263417}, "en": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "en", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 74444346, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 875142, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 433471, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 75752959, "size_in_bytes": 559716671}, "es": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "es", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 81383604, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 969821, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 478430, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 82831855, "size_in_bytes": 566795567}, "fr": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "fr", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 85809099, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 1029247, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 510112, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 87348458, "size_in_bytes": 571312170}, "hi": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "hi", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 170594284, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 2073081, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 1023923, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 173691288, "size_in_bytes": 657655000}, "ru": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "ru", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 129859935, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 1603474, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 786450, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 132249859, "size_in_bytes": 616213571}, "sw": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "sw", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 69286045, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 871659, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 429858, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 70587562, "size_in_bytes": 554551274}, "th": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "th", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 176063212, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 2147023, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 1061168, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 179271403, "size_in_bytes": 663235115}, "tr": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "tr", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 71637460, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 934942, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 459316, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 73031718, "size_in_bytes": 556995430}, "ur": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "ur", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 96441806, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 1416249, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 699960, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 98558015, "size_in_bytes": 582521727}, "vi": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "vi", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 101417750, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 1190225, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 590688, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 103198663, "size_in_bytes": 587162375}, "zh": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"dtype": "string", "id": null, "_type": "Value"}, "hypothesis": {"dtype": "string", "id": null, "_type": "Value"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "zh", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 72225161, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 777937, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 384859, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 73387957, "size_in_bytes": 557351669}, "all_languages": {"description": "XNLI is a subset of a few thousand examples from MNLI which has been translated\ninto a 14 different languages (some low-ish resource). As with MNLI, the goal is\nto predict textual entailment (does sentence A imply/contradict/neither sentence\nB) and is a classification task (given two sentences, predict one of three\nlabels).\n", "citation": "@InProceedings{conneau2018xnli,\n  author = {Conneau, Alexis\n                 and Rinott, Ruty\n                 and Lample, Guillaume\n                 and Williams, Adina\n                 and Bowman, Samuel R.\n                 and Schwenk, Holger\n                 and Stoyanov, Veselin},\n  title = {XNLI: Evaluating Cross-lingual Sentence Representations},\n  booktitle = {Proceedings of the 2018 Conference on Empirical Methods\n               in Natural Language Processing},\n  year = {2018},\n  publisher = {Association for Computational Linguistics},\n  location = {Brussels, Belgium},\n}", "homepage": "https://www.nyu.edu/projects/bowman/xnli/", "license": "", "features": {"premise": {"languages": ["ar", "bg", "de", "el", "en", "es", "fr", "hi", "ru", "sw", "th", "tr", "ur", "vi", "zh"], "id": null, "_type": "Translation"}, "hypothesis": {"languages": ["ar", "bg", "de", "el", "en", "es", "fr", "hi", "ru", "sw", "th", "tr", "ur", "vi", "zh"], "num_languages": 15, "id": null, "_type": "TranslationVariableLanguages"}, "label": {"num_classes": 3, "names": ["entailment", "neutral", "contradiction"], "names_file": null, "id": null, "_type": "ClassLabel"}}, "post_processed": null, "supervised_keys": null, "builder_name": "xnli", "config_name": "all_languages", "version": {"version_str": "1.1.0", "description": "", "major": 1, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1581474731, "num_examples": 392702, "dataset_name": "xnli"}, "test": {"name": "test", "num_bytes": 19387508, "num_examples": 5010, "dataset_name": "xnli"}, "validation": {"name": "validation", "num_bytes": 9566255, "num_examples": 2490, "dataset_name": "xnli"}}, "download_checksums": {"https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip": {"num_bytes": 466098360, "checksum": "f732517ba2fb1d550e9f3c2aabaef6017c91ee2dcec90e878f138764d224db05"}, "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip": {"num_bytes": 17865352, "checksum": "4ba1d5e1afdb7161f0f23c66dc787802ccfa8a25a3ddd3b165a35e50df346ab1"}}, "download_size": 483963712, "post_processing_size": null, "dataset_size": 1610428494, "size_in_bytes": 2094392206}}

dummy/{plain_text/1.0.0 → all_languages/1.1.0}/dummy_data.zip

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:087e5fb69a4751ad440cfac64e14928dd0ada6afe3d1ec2caaaaf2128b00da6b
-size 6032
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/ar/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/bg/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/de/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/el/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/en/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/es/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/fr/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/hi/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/ru/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/sw/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/th/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/tr/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/ur/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/vi/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

dummy/zh/1.1.0/dummy_data.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:72bfec36ff57afcd699fdfc3dd8020eb3830715d79ab98f89ddba9e6bd90bc32
+size 28285

xnli.py

@@ -21,8 +21,7 @@ from __future__ import absolute_import, division, print_function
 import collections
 import csv
 import os
-
-import six
+from contextlib import ExitStack
 
 import datasets
 
@@ -52,36 +51,76 @@ B) and is a classification task (given two sentences, predict one of three
 labels).
 """
 
-_DATA_URL = "https://cims.nyu.edu/~sbowman/xnli/XNLI-1.0.zip"
+_TRAIN_DATA_URL = "https://dl.fbaipublicfiles.com/XNLI/XNLI-MT-1.0.zip"
+_TESTVAL_DATA_URL = "https://dl.fbaipublicfiles.com/XNLI/XNLI-1.0.zip"
 
 _LANGUAGES = ("ar", "bg", "de", "el", "en", "es", "fr", "hi", "ru", "sw", "th", "tr", "ur", "vi", "zh")
 
 
+class XnliConfig(datasets.BuilderConfig):
+    """BuilderConfig for XNLI."""
+
+    def __init__(self, language: str, languages=None, **kwargs):
+        """BuilderConfig for XNLI.
+
+        Args:
+        language: One of ar,bg,de,el,en,es,fr,hi,ru,sw,th,tr,ur,vi,zh, or all_languages
+          **kwargs: keyword arguments forwarded to super.
+        """
+        super(XnliConfig, self).__init__(**kwargs)
+        self.language = language
+        if language != "all_languages":
+            self.languages = [language]
+        else:
+            self.languages = languages if languages is not None else _LANGUAGES
+
+
 class Xnli(datasets.GeneratorBasedBuilder):
     """XNLI: The Cross-Lingual NLI Corpus. Version 1.0."""
 
+    VERSION = datasets.Version("1.1.0", "")
+    BUILDER_CONFIG_CLASS = XnliConfig
     BUILDER_CONFIGS = [
-        datasets.BuilderConfig(
-            name="plain_text",
-            version=datasets.Version("1.0.0", ""),
-            description="Plain text import of XNLI",
+        XnliConfig(
+            name=lang,
+            language=lang,
+            version=datasets.Version("1.1.0", ""),
+            description=f"Plain text import of XNLI for the {lang} language",
+        )
+        for lang in _LANGUAGES
+    ] + [
+        XnliConfig(
+            name="all_languages",
+            language="all_languages",
+            version=datasets.Version("1.1.0", ""),
+            description="Plain text import of XNLI for all languages",
         )
     ]
 
     def _info(self):
-        return datasets.DatasetInfo(
-            description=_DESCRIPTION,
-            features=datasets.Features(
+        if self.config.language == "all_languages":
+            features = datasets.Features(
                 {
-                    "premise": datasets.features.Translation(
+                    "premise": datasets.Translation(
                         languages=_LANGUAGES,
                     ),
-                    "hypothesis": datasets.features.TranslationVariableLanguages(
+                    "hypothesis": datasets.TranslationVariableLanguages(
                         languages=_LANGUAGES,
                     ),
-                    "label": datasets.features.ClassLabel(names=["entailment", "neutral", "contradiction"]),
+                    "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
                 }
-            ),
+            )
+        else:
+            features = datasets.Features(
+                {
+                    "premise": datasets.Value("string"),
+                    "hypothesis": datasets.Value("string"),
+                    "label": datasets.ClassLabel(names=["entailment", "neutral", "contradiction"]),
+                }
+            )
+        return datasets.DatasetInfo(
+            description=_DESCRIPTION,
+            features=features,
             # No default supervised_keys (as we have to pass both premise
             # and hypothesis as input).
             supervised_keys=None,
@@ -90,31 +129,84 @@ class Xnli(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        dl_dir = dl_manager.download_and_extract(_DATA_URL)
-        data_dir = os.path.join(dl_dir, "XNLI-1.0")
+        dl_dirs = dl_manager.download_and_extract(
+            {
+                "train_data": _TRAIN_DATA_URL,
+                "testval_data": _TESTVAL_DATA_URL,
+            }
+        )
+        train_dir = os.path.join(dl_dirs["train_data"], "XNLI-MT-1.0", "multinli")
+        testval_dir = os.path.join(dl_dirs["testval_data"], "XNLI-1.0")
         return [
             datasets.SplitGenerator(
-                name=datasets.Split.TEST, gen_kwargs={"filepath": os.path.join(data_dir, "xnli.test.tsv")}
+                name=datasets.Split.TRAIN,
+                gen_kwargs={
+                    "filepaths": [
+                        os.path.join(train_dir, "multinli.train.{lang}.tsv".format(lang=lang))
+                        for lang in self.config.languages
+                    ],
+                    "data_format": "XNLI-MT",
+                },
             ),
             datasets.SplitGenerator(
-                name=datasets.Split.VALIDATION, gen_kwargs={"filepath": os.path.join(data_dir, "xnli.dev.tsv")}
+                name=datasets.Split.TEST,
+                gen_kwargs={"filepaths": [os.path.join(testval_dir, "xnli.test.tsv")], "data_format": "XNLI"},
+            ),
+            datasets.SplitGenerator(
+                name=datasets.Split.VALIDATION,
+                gen_kwargs={"filepaths": [os.path.join(testval_dir, "xnli.dev.tsv")], "data_format": "XNLI"},
             ),
         ]
 
-    def _generate_examples(self, filepath):
+    def _generate_examples(self, data_format, filepaths):
         """This function returns the examples in the raw (text) form."""
-        rows_per_pair_id = collections.defaultdict(list)
-
-        with open(filepath, encoding="utf-8") as f:
-            reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
-            for row in reader:
-                rows_per_pair_id[row["pairID"]].append(row)
-
-        for rows in six.itervalues(rows_per_pair_id):
-            premise = {row["language"]: row["sentence1"] for row in rows}
-            hypothesis = {row["language"]: row["sentence2"] for row in rows}
-            yield rows[0]["pairID"], {
-                "premise": premise,
-                "hypothesis": hypothesis,
-                "label": rows[0]["gold_label"],
-            }
+
+        if self.config.language == "all_languages":
+            if data_format == "XNLI-MT":
+                with ExitStack() as stack:
+                    files = [stack.enter_context(open(filepath, encoding="utf-8")) for filepath in filepaths]
+                    readers = [csv.DictReader(file, delimiter="\t", quoting=csv.QUOTE_NONE) for file in files]
+                    for row_idx, rows in enumerate(zip(*readers)):
+                        yield row_idx, {
+                            "premise": {lang: row["premise"] for lang, row in zip(self.config.languages, rows)},
+                            "hypothesis": {lang: row["hypo"] for lang, row in zip(self.config.languages, rows)},
+                            "label": rows[0]["label"].replace("contradictory", "contradiction"),
+                        }
+            else:
+                rows_per_pair_id = collections.defaultdict(list)
+                for filepath in filepaths:
+                    with open(filepath, encoding="utf-8") as f:
+                        reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
+                        for row in reader:
+                            rows_per_pair_id[row["pairID"]].append(row)
+
+                for rows in rows_per_pair_id.values():
+                    premise = {row["language"]: row["sentence1"] for row in rows}
+                    hypothesis = {row["language"]: row["sentence2"] for row in rows}
+                    yield rows[0]["pairID"], {
+                        "premise": premise,
+                        "hypothesis": hypothesis,
+                        "label": rows[0]["gold_label"],
+                    }
+        else:
+            if data_format == "XNLI-MT":
+                for file_idx, filepath in enumerate(filepaths):
+                    file = open(filepath, encoding="utf-8")
+                    reader = csv.DictReader(file, delimiter="\t", quoting=csv.QUOTE_NONE)
+                    for row_idx, row in enumerate(reader):
+                        yield (file_idx, row_idx), {
+                            "premise": row["premise"],
+                            "hypothesis": row["hypo"],
+                            "label": row["label"].replace("contradictory", "contradiction"),
+                        }
+            else:
+                for filepath in filepaths:
+                    with open(filepath, encoding="utf-8") as f:
+                        reader = csv.DictReader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
+                        for row in reader:
+                            if row["language"] == self.config.language:
+                                yield row["pairID"], {
+                                    "premise": row["sentence1"],
+                                    "hypothesis": row["sentence2"],
+                                    "label": row["gold_label"],
+                                }