{ "deu-ell" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 512218, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 238898, "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "name" : "test" }, "validation" : { "num_examples" : 2800, "dataset_name" : "tatoeba_mt", "num_bytes" : 273320, "name" : "validation" } }, "config_name" : "deu-ell", "dataset_size" : 5299, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 } }, "glg-por" : { "config_name" : "glg-por", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 432, "num_bytes" : 33423, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 33423, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 432 }, "isl-jpn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 21926, "config_name" : "isl-jpn", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 250, "num_bytes" : 21926, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 250, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-rus" : { "dataset_size" : 113121, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 11717570, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 1345955, "num_examples" : 12799, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 100322, "num_bytes" : 10371615, "name" : "validation" } }, "config_name" : "deu-rus" }, "fin-lat" : { "dataset_size" : 293, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_examples" : 293, "dataset_name" : "tatoeba_mt", "num_bytes" : 18696, "name" : "test" } }, "config_name" : "fin-lat", "size_in_bytes" : 18696, "builder_name" : "tatoeba_mt" }, "nld-nor" : { "size_in_bytes" : 15232, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 202, "dataset_name" : "tatoeba_mt", "num_bytes" : 15232 } }, "config_name" : "nld-nor", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 202, "supervised_keys" : null }, "epo-jpn" : { "dataset_size" : 9887, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_examples" : 3476, "dataset_name" : "tatoeba_mt", "num_bytes" : 334962 }, "validation" : { "name" : "validation", "num_bytes" : 614123, "dataset_name" : "tatoeba_mt", "num_examples" : 6411 } }, "config_name" : "epo-jpn", "size_in_bytes" : 949085, "builder_name" : "tatoeba_mt" }, "deu-kab" : { "splits" : { "test" : { "num_examples" : 372, "dataset_name" : "tatoeba_mt", "num_bytes" : 25648, "name" : "test" } }, "config_name" : "deu-kab", "size_in_bytes" : 25648, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 372, "supervised_keys" : null }, "heb-heb" : { "config_name" : "heb-heb", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 71416, "dataset_name" : "tatoeba_mt", "num_examples" : 731 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 999, "num_bytes" : 96379 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 167795, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1730 }, "cor-spa" : { "supervised_keys" : null, "dataset_size" : 205, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 11511, "config_name" : "cor-spa", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 205, "num_bytes" : 11511 } } }, "eng-hoc_Latn" : { "config_name" : "eng-hoc_Latn", "splits" : { "test" : { "num_bytes" : 36698, "num_examples" : 658, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 36698, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 658 }, "eng-cmn_Hant" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 23938, "config_name" : "eng-cmn_Hant", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 1579337, "dataset_name" : "tatoeba_mt", "num_examples" : 19463 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 4475, "num_bytes" : 360736, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1940073, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "epo-ido" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 113939, "config_name" : "epo-ido", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 416, "num_bytes" : 26630 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1182, "num_bytes" : 87309 } }, "supervised_keys" : null, "dataset_size" : 1598, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "fin-kor" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 33898, "config_name" : "fin-kor", "splits" : { "test" : { "name" : "test", "num_bytes" : 33898, "num_examples" : 421, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 421, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "heb-lad" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 217, "num_bytes" : 18893, "name" : "test" }, "validation" : { "num_bytes" : 3126, "num_examples" : 40, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "heb-lad", "size_in_bytes" : 22019, "builder_name" : "tatoeba_mt", "dataset_size" : 257, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "fin-jpn_Hira" : { "config_name" : "fin-jpn_Hira", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 913048, "dataset_name" : "tatoeba_mt", "num_examples" : 9437 }, "test" : { "name" : "test", "num_bytes" : 841840, "dataset_name" : "tatoeba_mt", "num_examples" : 8714 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1754888, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 18151 }, "aze-tur" : { "supervised_keys" : null, "dataset_size" : 1149, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 81567, "config_name" : "aze-tur", "splits" : { "test" : { "name" : "test", "num_bytes" : 81567, "num_examples" : 1149, "dataset_name" : "tatoeba_mt" } } }, "nds-por" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 12995, "num_examples" : 206, "dataset_name" : "tatoeba_mt" } }, "config_name" : "nds-por", "size_in_bytes" : 12995, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 206, "supervised_keys" : null }, "dtp-zsm_Latn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 52979, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 508, "num_bytes" : 52979 } }, "config_name" : "dtp-zsm_Latn", "dataset_size" : 508, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "nds-nld" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 2668, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1012, "num_bytes" : 70369 }, "test" : { "name" : "test", "num_bytes" : 117605, "dataset_name" : "tatoeba_mt", "num_examples" : 1656 } }, "config_name" : "nds-nld", "size_in_bytes" : 187974, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "arq-eng" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1139, "builder_name" : "tatoeba_mt", "size_in_bytes" : 104421, "config_name" : "arq-eng", "splits" : { "validation" : { "num_examples" : 735, "dataset_name" : "tatoeba_mt", "num_bytes" : 68363, "name" : "validation" }, "test" : { "num_examples" : 404, "dataset_name" : "tatoeba_mt", "num_bytes" : 36058, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ell-ita" : { "splits" : { "test" : { "num_bytes" : 36093, "dataset_name" : "tatoeba_mt", "num_examples" : 423, "name" : "test" } }, "config_name" : "ell-ita", "size_in_bytes" : 36093, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 423, "supervised_keys" : null }, "deu-lfn_Latn" : { "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-lfn_Latn", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2627, "num_examples" : 31, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 25836, "dataset_name" : "tatoeba_mt", "num_examples" : 290 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 28463, "supervised_keys" : null, "dataset_size" : 321, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "orv-ukr" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 95946, "config_name" : "orv-ukr", "splits" : { "test" : { "num_bytes" : 95946, "num_examples" : 972, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 972 }, "hun-spa" : { "supervised_keys" : null, "dataset_size" : 6692, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 483123, "config_name" : "hun-spa", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 179662, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 303461, "num_examples" : 4193, "dataset_name" : "tatoeba_mt" } } }, "ita-cmn_Hans" : { "dataset_size" : 2406, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 138936, "dataset_name" : "tatoeba_mt", "num_examples" : 1826 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 580, "num_bytes" : 45045, "name" : "validation" } }, "config_name" : "ita-cmn_Hans", "size_in_bytes" : 183981, "builder_name" : "tatoeba_mt" }, "bre-fra" : { "supervised_keys" : null, "dataset_size" : 5552, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 328164, "config_name" : "bre-fra", "splits" : { "test" : { "name" : "test", "num_bytes" : 145228, "dataset_name" : "tatoeba_mt", "num_examples" : 2493 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 3059, "num_bytes" : 182936, "name" : "validation" } } }, "msa-zho" : { "supervised_keys" : null, "dataset_size" : 371, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 33792, "config_name" : "msa-zho", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 371, "num_bytes" : 33792, "name" : "test" } } }, "epo-ile_Latn" : { "splits" : { "test" : { "name" : "test", "num_examples" : 314, "dataset_name" : "tatoeba_mt", "num_bytes" : 19423 } }, "config_name" : "epo-ile_Latn", "size_in_bytes" : 19423, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 314, "supervised_keys" : null }, "heb-lfn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 30410, "config_name" : "heb-lfn", "splits" : { "test" : { "num_bytes" : 24022, "num_examples" : 279, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 71, "num_bytes" : 6388, "name" : "validation" } }, "supervised_keys" : null, "dataset_size" : 350, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0" }, "bul-cmn_Hans" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 277, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 31599, "dataset_name" : "tatoeba_mt", "num_examples" : 277, "name" : "test" } }, "config_name" : "bul-cmn_Hans", "size_in_bytes" : 31599, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "deu-srp_Latn" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 985, "supervised_keys" : null, "size_in_bytes" : 94067, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 985, "dataset_name" : "tatoeba_mt", "num_bytes" : 94067 } }, "config_name" : "deu-srp_Latn", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } } }, "deu-nds" : { "supervised_keys" : null, "dataset_size" : 17840, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-nds", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 7842, "num_bytes" : 630733, "name" : "validation" }, "test" : { "num_examples" : 9998, "dataset_name" : "tatoeba_mt", "num_bytes" : 796333, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1427066 }, "nld-pol" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "nld-pol", "splits" : { "validation" : { "num_examples" : 1005, "dataset_name" : "tatoeba_mt", "num_bytes" : 73045, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1192, "num_bytes" : 87695 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 160740, "supervised_keys" : null, "dataset_size" : 2197, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "deu-zho" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 901433, "config_name" : "deu-zho", "splits" : { "validation" : { "num_bytes" : 574107, "dataset_name" : "tatoeba_mt", "num_examples" : 6837, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 3943, "dataset_name" : "tatoeba_mt", "num_bytes" : 327326 } }, "supervised_keys" : null, "dataset_size" : 10780, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-kha" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 1332, "supervised_keys" : null, "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1313, "num_bytes" : 84084, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 19, "num_bytes" : 1635, "name" : "validation" } }, "config_name" : "eng-kha", "size_in_bytes" : 85719, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "pol-swe" : { "supervised_keys" : null, "dataset_size" : 1391, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 100023, "config_name" : "pol-swe", "splits" : { "test" : { "num_bytes" : 100023, "num_examples" : 1391, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "dan-nld" : { "supervised_keys" : null, "dataset_size" : 1677, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "dan-nld", "splits" : { "test" : { "num_examples" : 1642, "dataset_name" : "tatoeba_mt", "num_bytes" : 109243, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 35, "num_bytes" : 2485 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 111728 }, "ara-tur" : { "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "ara-tur", "splits" : { "test" : { "num_examples" : 1262, "dataset_name" : "tatoeba_mt", "num_bytes" : 109267, "name" : "test" }, "validation" : { "num_examples" : 51, "dataset_name" : "tatoeba_mt", "num_bytes" : 4914, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 114181, "supervised_keys" : null, "dataset_size" : 1313, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "heb-cmn_Hans" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 22756, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 284, "num_bytes" : 22756, "name" : "test" } }, "config_name" : "heb-cmn_Hans", "dataset_size" : 284, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "dan-por" : { "supervised_keys" : null, "dataset_size" : 872, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 54721, "config_name" : "dan-por", "splits" : { "test" : { "name" : "test", "num_examples" : 872, "dataset_name" : "tatoeba_mt", "num_bytes" : 54721 } } }, "deu-hrx" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-hrx", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 470, "num_bytes" : 32127, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 32127, "supervised_keys" : null, "dataset_size" : 470, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "lat_Latn-lfn_Latn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 22360, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 3258, "dataset_name" : "tatoeba_mt", "num_examples" : 49, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 19102, "num_examples" : 298, "dataset_name" : "tatoeba_mt" } }, "config_name" : "lat_Latn-lfn_Latn", "dataset_size" : 347, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 } }, "jpn_Hira-nob" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 83793, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 804, "num_bytes" : 83793, "name" : "test" } }, "config_name" : "jpn_Hira-nob", "dataset_size" : 804, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "fin-kur" : { "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 258, "builder_name" : "tatoeba_mt", "size_in_bytes" : 20711, "config_name" : "fin-kur", "splits" : { "test" : { "name" : "test", "num_bytes" : 20711, "dataset_name" : "tatoeba_mt", "num_examples" : 258 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "hbs-jpn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 35763, "config_name" : "hbs-jpn", "splits" : { "test" : { "name" : "test", "num_bytes" : 35763, "num_examples" : 399, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 399, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-yue_Hans" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 322542, "config_name" : "eng-yue_Hans", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 2719, "num_bytes" : 258827 }, "test" : { "num_examples" : 676, "dataset_name" : "tatoeba_mt", "num_bytes" : 63715, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 3395 }, "epo-swe" : { "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1823, "builder_name" : "tatoeba_mt", "size_in_bytes" : 124975, "config_name" : "epo-swe", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 5259, "num_examples" : 67, "dataset_name" : "tatoeba_mt" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1756, "num_bytes" : 119716, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "slv-ukr" : { "dataset_size" : 962, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 914, "num_bytes" : 60177 }, "validation" : { "num_bytes" : 3873, "num_examples" : 48, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "slv-ukr", "size_in_bytes" : 64050, "builder_name" : "tatoeba_mt" }, "dan-fin" : { "dataset_size" : 4406, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 178869, "dataset_name" : "tatoeba_mt", "num_examples" : 2664 }, "validation" : { "name" : "validation", "num_bytes" : 125290, "dataset_name" : "tatoeba_mt", "num_examples" : 1742 } }, "config_name" : "dan-fin", "size_in_bytes" : 304159, "builder_name" : "tatoeba_mt" }, "lad_Latn-lfn_Latn" : { "dataset_size" : 238, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 15333, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 210, "num_bytes" : 13412, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 28, "num_bytes" : 1921, "name" : "validation" } }, "config_name" : "lad_Latn-lfn_Latn" }, "swe-cmn_Hans" : { "dataset_size" : 477, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 33354, "dataset_name" : "tatoeba_mt", "num_examples" : 477 } }, "config_name" : "swe-cmn_Hans", "size_in_bytes" : 33354, "builder_name" : "tatoeba_mt" }, "jpn_Hira-cmn_Hant" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 141240, "config_name" : "jpn_Hira-cmn_Hant", "splits" : { "test" : { "name" : "test", "num_bytes" : 65875, "dataset_name" : "tatoeba_mt", "num_examples" : 681 }, "validation" : { "num_bytes" : 75365, "num_examples" : 778, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1459 }, "jpn_Hira-kor_Hang" : { "dataset_size" : 511, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_examples" : 511, "dataset_name" : "tatoeba_mt", "num_bytes" : 53616 } }, "config_name" : "jpn_Hira-kor_Hang", "size_in_bytes" : 53616, "builder_name" : "tatoeba_mt" }, "fra-wuu" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 107328, "config_name" : "fra-wuu", "splits" : { "test" : { "num_examples" : 744, "dataset_name" : "tatoeba_mt", "num_bytes" : 65808, "name" : "test" }, "validation" : { "num_examples" : 493, "dataset_name" : "tatoeba_mt", "num_bytes" : 41520, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1237 }, "eng-got" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 20327, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 20327, "dataset_name" : "tatoeba_mt", "num_examples" : 201 } }, "config_name" : "eng-got", "dataset_size" : 201, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "por-ukr" : { "supervised_keys" : null, "dataset_size" : 4395, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 372859, "config_name" : "por-ukr", "splits" : { "test" : { "num_bytes" : 285017, "dataset_name" : "tatoeba_mt", "num_examples" : 3371, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1024, "num_bytes" : 87842, "name" : "validation" } } }, "lit-tur" : { "supervised_keys" : null, "dataset_size" : 1471, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "lit-tur", "splits" : { "test" : { "name" : "test", "num_bytes" : 102387, "num_examples" : 1471, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 102387 }, "lat-rus" : { "supervised_keys" : null, "dataset_size" : 2074, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 191405, "config_name" : "lat-rus", "splits" : { "validation" : { "num_bytes" : 95812, "dataset_name" : "tatoeba_mt", "num_examples" : 1034, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 95593, "num_examples" : 1040, "dataset_name" : "tatoeba_mt" } } }, "epo-slv" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 23153, "config_name" : "epo-slv", "splits" : { "test" : { "num_bytes" : 23153, "dataset_name" : "tatoeba_mt", "num_examples" : 301, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 301 }, "cat-eng" : { "dataset_size" : 1857, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 150171, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 126706, "num_examples" : 1630, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_examples" : 227, "dataset_name" : "tatoeba_mt", "num_bytes" : 23465, "name" : "validation" } }, "config_name" : "cat-eng" }, "ara-heb" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 119681, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 110757, "dataset_name" : "tatoeba_mt", "num_examples" : 1207, "name" : "test" }, "validation" : { "num_bytes" : 8924, "dataset_name" : "tatoeba_mt", "num_examples" : 90, "name" : "validation" } }, "config_name" : "ara-heb", "dataset_size" : 1297, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "dan-jpn" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 74725, "config_name" : "dan-jpn", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 903, "num_bytes" : 74725 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 903 }, "deu-ron" : { "config_name" : "deu-ron", "splits" : { "test" : { "num_bytes" : 84604, "num_examples" : 1140, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1012, "num_bytes" : 74503, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 159107, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 2152 }, "fra-tgl" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 835, "config_name" : "fra-tgl", "splits" : { "test" : { "num_bytes" : 69627, "dataset_name" : "tatoeba_mt", "num_examples" : 835, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 69627, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-ind" : { "size_in_bytes" : 780481, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 450871, "num_examples" : 5808, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 4288, "num_bytes" : 329610 } }, "config_name" : "eng-ind", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 10096, "supervised_keys" : null }, "rus-xal" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 208, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 23410, "num_examples" : 208, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "rus-xal", "size_in_bytes" : 23410, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-kat" : { "splits" : { "test" : { "num_examples" : 983, "dataset_name" : "tatoeba_mt", "num_bytes" : 94570, "name" : "test" } }, "config_name" : "eng-kat", "size_in_bytes" : 94570, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 983, "supervised_keys" : null }, "zsm_Latn-ind" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 224, "config_name" : "zsm_Latn-ind", "splits" : { "test" : { "name" : "test", "num_bytes" : 23540, "dataset_name" : "tatoeba_mt", "num_examples" : 224 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 23540, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "spa-vie" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 593, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_bytes" : 46565, "dataset_name" : "tatoeba_mt", "num_examples" : 593 } }, "config_name" : "spa-vie", "size_in_bytes" : 46565, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "heb-hun" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 32411, "config_name" : "heb-hun", "splits" : { "test" : { "num_bytes" : 32411, "num_examples" : 416, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 416 }, "ita-tur" : { "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 9999, "num_bytes" : 899367 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 5702, "num_bytes" : 507078 } }, "config_name" : "ita-tur", "size_in_bytes" : 1406445, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 15701, "supervised_keys" : null }, "srp_Cyrl-ukr" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 19037, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 19037, "dataset_name" : "tatoeba_mt", "num_examples" : 204, "name" : "test" } }, "config_name" : "srp_Cyrl-ukr", "dataset_size" : 204, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 } }, "ita-toki" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_examples" : 199, "dataset_name" : "tatoeba_mt", "num_bytes" : 17073 } }, "config_name" : "ita-toki", "size_in_bytes" : 17073, "builder_name" : "tatoeba_mt", "dataset_size" : 199, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "eng-tzl_Latn" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 199, "config_name" : "eng-tzl_Latn", "splits" : { "test" : { "num_bytes" : 10349, "dataset_name" : "tatoeba_mt", "num_examples" : 199, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 10349, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fin-nno" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 50880, "dataset_name" : "tatoeba_mt", "num_examples" : 815 }, "validation" : { "name" : "validation", "num_bytes" : 26221, "dataset_name" : "tatoeba_mt", "num_examples" : 400 } }, "config_name" : "fin-nno", "size_in_bytes" : 77101, "builder_name" : "tatoeba_mt", "dataset_size" : 1215, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "fra-msa" : { "config_name" : "fra-msa", "splits" : { "test" : { "name" : "test", "num_bytes" : 85373, "num_examples" : 1002, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 85373, "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1002 }, "jpn-tur" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 78766, "dataset_name" : "tatoeba_mt", "num_examples" : 810 } }, "config_name" : "jpn-tur", "size_in_bytes" : 78766, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 810, "supervised_keys" : null }, "ces-spa" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 145230, "config_name" : "ces-spa", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 133, "num_bytes" : 8887 }, "test" : { "name" : "test", "num_examples" : 1806, "dataset_name" : "tatoeba_mt", "num_bytes" : 136343 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1939 }, "kaz_Cyrl-rus" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 337316, "num_examples" : 1397, "dataset_name" : "tatoeba_mt" } }, "config_name" : "kaz_Cyrl-rus", "size_in_bytes" : 337316, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 1397, "supervised_keys" : null }, "eng-prg" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 19065, "dataset_name" : "tatoeba_mt", "num_examples" : 212 } }, "config_name" : "eng-prg", "size_in_bytes" : 19065, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 212, "supervised_keys" : null }, "epo-por" : { "size_in_bytes" : 5365783, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 10808, "num_bytes" : 1038541 }, "validation" : { "name" : "validation", "num_examples" : 46092, "dataset_name" : "tatoeba_mt", "num_bytes" : 4327242 } }, "config_name" : "epo-por", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 56900, "supervised_keys" : null }, "heb-tur" : { "supervised_keys" : null, "dataset_size" : 1405, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "heb-tur", "splits" : { "validation" : { "num_examples" : 29, "dataset_name" : "tatoeba_mt", "num_bytes" : 1985, "name" : "validation" }, "test" : { "num_bytes" : 104779, "num_examples" : 1376, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 106764 }, "ell-fra" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1505, "config_name" : "ell-fra", "splits" : { "test" : { "num_bytes" : 135841, "dataset_name" : "tatoeba_mt", "num_examples" : 1505, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 135841, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "epo-nld" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 93676, "builder_name" : "tatoeba_mt", "size_in_bytes" : 7509210, "config_name" : "epo-nld", "splits" : { "test" : { "num_bytes" : 1024905, "dataset_name" : "tatoeba_mt", "num_examples" : 12840, "name" : "test" }, "validation" : { "num_bytes" : 6484305, "dataset_name" : "tatoeba_mt", "num_examples" : 80836, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "fra-oci" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 68132, "config_name" : "fra-oci", "splits" : { "test" : { "name" : "test", "num_examples" : 805, "dataset_name" : "tatoeba_mt", "num_bytes" : 68132 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 805 }, "epo-vol" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 57394, "config_name" : "epo-vol", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2338, "num_examples" : 36, "dataset_name" : "tatoeba_mt" }, "test" : { "num_examples" : 818, "dataset_name" : "tatoeba_mt", "num_bytes" : 55056, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 854 }, "epo-hrv" : { "config_name" : "epo-hrv", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 474, "num_bytes" : 35993 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 422, "num_bytes" : 33262 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 69255, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 896 }, "jpn-toki" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 241, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_examples" : 241, "dataset_name" : "tatoeba_mt", "num_bytes" : 20860 } }, "config_name" : "jpn-toki", "size_in_bytes" : 20860, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "rus-spa" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 9801059, "config_name" : "rus-spa", "splits" : { "validation" : { "num_bytes" : 8750687, "num_examples" : 86902, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 1050372, "num_examples" : 10505, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 97407, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-urd" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1662, "builder_name" : "tatoeba_mt", "size_in_bytes" : 160779, "config_name" : "eng-urd", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1662, "num_bytes" : 160779 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "deu-run" : { "dataset_size" : 2941, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 174984, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 1190, "dataset_name" : "tatoeba_mt", "num_bytes" : 74444, "name" : "validation" }, "test" : { "num_examples" : 1751, "dataset_name" : "tatoeba_mt", "num_bytes" : 100540, "name" : "test" } }, "config_name" : "deu-run" }, "fra-uig_Arab" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 692, "builder_name" : "tatoeba_mt", "size_in_bytes" : 66689, "config_name" : "fra-uig_Arab", "splits" : { "test" : { "num_bytes" : 66689, "dataset_name" : "tatoeba_mt", "num_examples" : 692, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ber_Latn-deu" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 658, "num_bytes" : 55766, "name" : "test" } }, "config_name" : "ber_Latn-deu", "size_in_bytes" : 55766, "builder_name" : "tatoeba_mt", "dataset_size" : 658, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "fin-jpn_Kana" : { "supervised_keys" : null, "dataset_size" : 566, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 52953, "config_name" : "fin-jpn_Kana", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 225, "num_bytes" : 20953 }, "validation" : { "name" : "validation", "num_bytes" : 32000, "dataset_name" : "tatoeba_mt", "num_examples" : 341 } } }, "ara-ber_Latn" : { "dataset_size" : 3322, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 208463, "dataset_name" : "tatoeba_mt", "num_examples" : 2298, "name" : "test" }, "validation" : { "num_bytes" : 94416, "dataset_name" : "tatoeba_mt", "num_examples" : 1024, "name" : "validation" } }, "config_name" : "ara-ber_Latn", "size_in_bytes" : 302879, "builder_name" : "tatoeba_mt" }, "bul-tur" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 88651, "dataset_name" : "tatoeba_mt", "num_examples" : 833 } }, "config_name" : "bul-tur", "size_in_bytes" : 88651, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 833, "supervised_keys" : null }, "afr-epo" : { "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "afr-epo", "splits" : { "test" : { "num_bytes" : 81498, "dataset_name" : "tatoeba_mt", "num_examples" : 1115, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 81498, "supervised_keys" : null, "dataset_size" : 1115, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "chm-rus" : { "size_in_bytes" : 515945, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_examples" : 999, "dataset_name" : "tatoeba_mt", "num_bytes" : 133979 }, "test" : { "name" : "test", "num_bytes" : 381966, "num_examples" : 2749, "dataset_name" : "tatoeba_mt" } }, "config_name" : "chm-rus", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 3748, "supervised_keys" : null }, "ido-ina" : { "supervised_keys" : null, "dataset_size" : 366, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 24379, "config_name" : "ido-ina", "splits" : { "validation" : { "num_bytes" : 3612, "num_examples" : 51, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 20767, "num_examples" : 315, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "pol-por" : { "dataset_size" : 704, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 54377, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 704, "num_bytes" : 54377, "name" : "test" } }, "config_name" : "pol-por" }, "fra-spa" : { "dataset_size" : 50887, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 40594, "num_bytes" : 3483005 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 10293, "num_bytes" : 883143 } }, "config_name" : "fra-spa", "size_in_bytes" : 4366148, "builder_name" : "tatoeba_mt" }, "lav-rus" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 22259, "num_examples" : 273, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "lav-rus", "size_in_bytes" : 22259, "builder_name" : "tatoeba_mt", "dataset_size" : 273, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "rus-uig_Arab" : { "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 528, "num_bytes" : 65448 } }, "config_name" : "rus-uig_Arab", "size_in_bytes" : 65448, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 528, "supervised_keys" : null }, "hsb-slv" : { "supervised_keys" : null, "dataset_size" : 742, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 35522, "config_name" : "hsb-slv", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 742, "num_bytes" : 35522, "name" : "test" } } }, "swe-tur" : { "dataset_size" : 202, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 14122, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 202, "num_bytes" : 14122 } }, "config_name" : "swe-tur" }, "epo-fin" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 282139, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 2861, "dataset_name" : "tatoeba_mt", "num_bytes" : 206658 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1046, "num_bytes" : 75481 } }, "config_name" : "epo-fin", "dataset_size" : 3907, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "dan-swe" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 205830, "config_name" : "dan-swe", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 98464, "num_examples" : 1439, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 107366, "num_examples" : 1548, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 2987, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "spa-tlh" : { "config_name" : "spa-tlh", "splits" : { "test" : { "num_bytes" : 20288, "dataset_name" : "tatoeba_mt", "num_examples" : 324, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 20288, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 324 }, "ido-yid" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 482, "supervised_keys" : null, "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 57, "num_bytes" : 5495, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 37649, "dataset_name" : "tatoeba_mt", "num_examples" : 425 } }, "config_name" : "ido-yid", "size_in_bytes" : 43144, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ita-msa" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 425, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 32292, "num_examples" : 425, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "ita-msa", "size_in_bytes" : 32292, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "dsb-hsb" : { "size_in_bytes" : 15387, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 15387, "num_examples" : 330, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "dsb-hsb", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 330, "supervised_keys" : null }, "fra-hun" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 4847, "num_bytes" : 345870 }, "test" : { "num_bytes" : 176603, "num_examples" : 2493, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "fra-hun", "size_in_bytes" : 522473, "builder_name" : "tatoeba_mt", "dataset_size" : 7340, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "eng-zho" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 53463, "config_name" : "eng-zho", "splits" : { "validation" : { "name" : "validation", "num_examples" : 43074, "dataset_name" : "tatoeba_mt", "num_bytes" : 3622494 }, "test" : { "num_bytes" : 871802, "num_examples" : 10389, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 4494296, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-nds" : { "dataset_size" : 5719, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 411144, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 232162, "num_examples" : 3220, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 178982 } }, "config_name" : "eng-nds" }, "epo-fas" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 1054715, "config_name" : "epo-fas", "splits" : { "test" : { "num_bytes" : 268962, "dataset_name" : "tatoeba_mt", "num_examples" : 2513, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 7429, "num_bytes" : 785753, "name" : "validation" } }, "supervised_keys" : null, "dataset_size" : 9942, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-srp_Latn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 808496, "config_name" : "eng-srp_Latn", "splits" : { "validation" : { "num_bytes" : 311488, "dataset_name" : "tatoeba_mt", "num_examples" : 4239, "name" : "validation" }, "test" : { "num_bytes" : 497008, "dataset_name" : "tatoeba_mt", "num_examples" : 6655, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 10894, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-gla" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 68841, "config_name" : "eng-gla", "splits" : { "test" : { "name" : "test", "num_bytes" : 68841, "num_examples" : 954, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 954 }, "jpn_Hani-spa" : { "dataset_size" : 3562, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 328542, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 2464, "num_bytes" : 226500 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1098, "num_bytes" : 102042, "name" : "test" } }, "config_name" : "jpn_Hani-spa" }, "eng-hrx" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 220, "builder_name" : "tatoeba_mt", "size_in_bytes" : 13943, "config_name" : "eng-hrx", "splits" : { "test" : { "num_bytes" : 13943, "dataset_name" : "tatoeba_mt", "num_examples" : 220, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "fin-fkv" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 455, "config_name" : "fin-fkv", "splits" : { "validation" : { "name" : "validation", "num_examples" : 67, "dataset_name" : "tatoeba_mt", "num_bytes" : 5280 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 388, "num_bytes" : 32367 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 37647, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "lad_Latn-yid" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "lad_Latn-yid", "splits" : { "test" : { "name" : "test", "num_bytes" : 37993, "dataset_name" : "tatoeba_mt", "num_examples" : 437 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 48, "num_bytes" : 4412, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 42405, "supervised_keys" : null, "dataset_size" : 485, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0" }, "ind-zsm_Latn" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 224, "num_bytes" : 23472, "name" : "test" } }, "config_name" : "ind-zsm_Latn", "size_in_bytes" : 23472, "builder_name" : "tatoeba_mt", "dataset_size" : 224, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 } }, "ina_Latn-lfn_Latn" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 55259, "config_name" : "ina_Latn-lfn_Latn", "splits" : { "validation" : { "num_examples" : 95, "dataset_name" : "tatoeba_mt", "num_bytes" : 6775, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 48484, "num_examples" : 691, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 786 }, "ces-hun" : { "supervised_keys" : null, "dataset_size" : 2023, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 128323, "config_name" : "ces-hun", "splits" : { "validation" : { "num_bytes" : 6685, "num_examples" : 113, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 121638, "num_examples" : 1910, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "ita-spa" : { "config_name" : "ita-spa", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 9246, "num_bytes" : 705745, "name" : "validation" }, "test" : { "num_examples" : 4999, "dataset_name" : "tatoeba_mt", "num_bytes" : 381788, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1087533, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 14245 }, "epo-pol" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 629898, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 439099, "dataset_name" : "tatoeba_mt", "num_examples" : 5785, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 190799, "num_examples" : 2512, "dataset_name" : "tatoeba_mt" } }, "config_name" : "epo-pol", "dataset_size" : 8297, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "cat-epo" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 771, "builder_name" : "tatoeba_mt", "size_in_bytes" : 61422, "config_name" : "cat-epo", "splits" : { "test" : { "num_bytes" : 61422, "dataset_name" : "tatoeba_mt", "num_examples" : 771, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "hbs-nor" : { "supervised_keys" : null, "dataset_size" : 11147, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "hbs-nor", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 575038, "dataset_name" : "tatoeba_mt", "num_examples" : 6148 }, "test" : { "name" : "test", "num_bytes" : 481328, "num_examples" : 4999, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1056366 }, "hun-cmn_Hans" : { "config_name" : "hun-cmn_Hans", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 246, "num_bytes" : 19862, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 19862, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 246 }, "dan-ita" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "dan-ita", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 283, "num_bytes" : 23819, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 23819, "supervised_keys" : null, "dataset_size" : 283, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "cat-ukr" : { "dataset_size" : 455, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 35263, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 35263, "dataset_name" : "tatoeba_mt", "num_examples" : 455, "name" : "test" } }, "config_name" : "cat-ukr" }, "jpn_Hira-ind" : { "supervised_keys" : null, "dataset_size" : 3128, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 290136, "config_name" : "jpn_Hira-ind", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2255, "num_bytes" : 209281, "name" : "test" }, "validation" : { "num_bytes" : 80855, "dataset_name" : "tatoeba_mt", "num_examples" : 873, "name" : "validation" } } }, "ina-rus" : { "dataset_size" : 1490, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_examples" : 38, "dataset_name" : "tatoeba_mt", "num_bytes" : 2799 }, "test" : { "name" : "test", "num_bytes" : 175954, "dataset_name" : "tatoeba_mt", "num_examples" : 1452 } }, "config_name" : "ina-rus", "size_in_bytes" : 178753, "builder_name" : "tatoeba_mt" }, "kzj-msa" : { "supervised_keys" : null, "dataset_size" : 369, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "kzj-msa", "splits" : { "test" : { "num_examples" : 369, "dataset_name" : "tatoeba_mt", "num_bytes" : 45722, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 45722 }, "uig-zho" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "uig-zho", "splits" : { "test" : { "name" : "test", "num_bytes" : 207018, "dataset_name" : "tatoeba_mt", "num_examples" : 1928 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 207018, "supervised_keys" : null, "dataset_size" : 1928, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-kab" : { "supervised_keys" : null, "dataset_size" : 26579, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-kab", "splits" : { "test" : { "name" : "test", "num_examples" : 12141, "dataset_name" : "tatoeba_mt", "num_bytes" : 812807 }, "validation" : { "num_examples" : 14438, "dataset_name" : "tatoeba_mt", "num_bytes" : 926831, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1739638 }, "jpn-spa" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 34160, "supervised_keys" : null, "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10741, "num_bytes" : 1065342, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 23419, "num_bytes" : 2334150 } }, "config_name" : "jpn-spa", "size_in_bytes" : 3399492, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-lad_Latn" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 40637, "num_examples" : 671, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "num_bytes" : 2197, "dataset_name" : "tatoeba_mt", "num_examples" : 36, "name" : "validation" } }, "config_name" : "eng-lad_Latn", "size_in_bytes" : 42834, "builder_name" : "tatoeba_mt", "dataset_size" : 707, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "rus-toki" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 89316, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2708, "num_examples" : 19, "dataset_name" : "tatoeba_mt" }, "test" : { "num_bytes" : 86608, "num_examples" : 899, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "rus-toki", "dataset_size" : 918, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "eng-rus" : { "splits" : { "test" : { "num_bytes" : 1947659, "dataset_name" : "tatoeba_mt", "num_examples" : 19424, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 48671339, "num_examples" : 504173, "dataset_name" : "tatoeba_mt" } }, "config_name" : "eng-rus", "size_in_bytes" : 50618998, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 523597, "supervised_keys" : null }, "yid-yid" : { "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 291, "num_bytes" : 35801, "name" : "test" } }, "config_name" : "yid-yid", "size_in_bytes" : 35801, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 291, "supervised_keys" : null }, "fra-nob" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 322, "config_name" : "fra-nob", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 322, "num_bytes" : 24303 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 24303, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ell-por" : { "size_in_bytes" : 73981, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 1845, "dataset_name" : "tatoeba_mt", "num_examples" : 22, "name" : "validation" }, "test" : { "num_bytes" : 72136, "num_examples" : 884, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "ell-por", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 906, "supervised_keys" : null }, "heb-spa" : { "supervised_keys" : null, "dataset_size" : 2924, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 273925, "config_name" : "heb-spa", "splits" : { "test" : { "num_bytes" : 174876, "dataset_name" : "tatoeba_mt", "num_examples" : 1848, "name" : "test" }, "validation" : { "num_examples" : 1076, "dataset_name" : "tatoeba_mt", "num_bytes" : 99049, "name" : "validation" } } }, "ell-nld" : { "splits" : { "test" : { "num_bytes" : 121339, "num_examples" : 921, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "num_bytes" : 9164, "dataset_name" : "tatoeba_mt", "num_examples" : 66, "name" : "validation" } }, "config_name" : "ell-nld", "size_in_bytes" : 130503, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 987, "supervised_keys" : null }, "epo-fra" : { "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "epo-fra", "splits" : { "test" : { "num_bytes" : 1084431, "dataset_name" : "tatoeba_mt", "num_examples" : 12167, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 287547, "dataset_name" : "tatoeba_mt", "num_bytes" : 25412672 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 26497103, "supervised_keys" : null, "dataset_size" : 299714, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "jbo_Latn-cmn_Hant" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 230, "num_bytes" : 15670, "name" : "test" }, "validation" : { "num_bytes" : 1549, "num_examples" : 20, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "jbo_Latn-cmn_Hant", "size_in_bytes" : 17219, "builder_name" : "tatoeba_mt", "dataset_size" : 250, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "rus-tur" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "rus-tur", "splits" : { "validation" : { "num_bytes" : 658089, "dataset_name" : "tatoeba_mt", "num_examples" : 6256, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 4999, "num_bytes" : 517450 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1175539, "supervised_keys" : null, "dataset_size" : 11255, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "bul-spa" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 24507, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 285, "num_bytes" : 24507, "name" : "test" } }, "config_name" : "bul-spa", "dataset_size" : 285, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "lfn-yid" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 118544, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 24927, "num_examples" : 243, "dataset_name" : "tatoeba_mt" }, "test" : { "num_bytes" : 93617, "num_examples" : 992, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "lfn-yid", "dataset_size" : 1235, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "kat-rus" : { "supervised_keys" : null, "dataset_size" : 641, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "kat-rus", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 641, "num_bytes" : 91485, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 91485 }, "deu-cmn_Hant" : { "dataset_size" : 3491, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 265327, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 2250, "dataset_name" : "tatoeba_mt", "num_bytes" : 172171, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 93156, "dataset_name" : "tatoeba_mt", "num_examples" : 1241 } }, "config_name" : "deu-cmn_Hant" }, "afr-eng" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 166348, "config_name" : "afr-eng", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1006, "num_bytes" : 69888 }, "test" : { "num_bytes" : 96460, "num_examples" : 1373, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2379 }, "eng-lfn_Latn" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 4934, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2449, "num_bytes" : 183672 }, "validation" : { "name" : "validation", "num_bytes" : 198066, "num_examples" : 2485, "dataset_name" : "tatoeba_mt" } }, "config_name" : "eng-lfn_Latn", "size_in_bytes" : 381738, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "epo-nor" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 4167, "supervised_keys" : null, "size_in_bytes" : 413322, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 3157, "num_bytes" : 313130, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 100192, "num_examples" : 1010, "dataset_name" : "tatoeba_mt" } }, "config_name" : "epo-nor", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "bel-spa" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 204, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_examples" : 204, "dataset_name" : "tatoeba_mt", "num_bytes" : 18897 } }, "config_name" : "bel-spa", "size_in_bytes" : 18897, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fra-tur" : { "supervised_keys" : null, "dataset_size" : 9298, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 752309, "config_name" : "fra-tur", "splits" : { "validation" : { "name" : "validation", "num_examples" : 6717, "dataset_name" : "tatoeba_mt", "num_bytes" : 543390 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2581, "num_bytes" : 208919 } } }, "jpn-msa" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "num_bytes" : 98191, "dataset_name" : "tatoeba_mt", "num_examples" : 1059, "name" : "validation" }, "test" : { "num_bytes" : 242969, "num_examples" : 2615, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "jpn-msa", "size_in_bytes" : 341160, "builder_name" : "tatoeba_mt", "dataset_size" : 3674, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 } }, "eng-zza" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 528, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 29734, "num_examples" : 528, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "eng-zza", "size_in_bytes" : 29734, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "nob-spa" : { "splits" : { "test" : { "name" : "test", "num_examples" : 884, "dataset_name" : "tatoeba_mt", "num_bytes" : 74321 } }, "config_name" : "nob-spa", "size_in_bytes" : 74321, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 884, "supervised_keys" : null }, "dan-deu" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 16917, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1390051, "config_name" : "dan-deu", "splits" : { "test" : { "name" : "test", "num_bytes" : 819673, "num_examples" : 9997, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 570378, "dataset_name" : "tatoeba_mt", "num_examples" : 6920, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "jpn_Hira-ukr" : { "dataset_size" : 382, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 41595, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 41595, "num_examples" : 382, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "jpn_Hira-ukr" }, "fra-toki" : { "dataset_size" : 586, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 51475, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_examples" : 25, "dataset_name" : "tatoeba_mt", "num_bytes" : 3337 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 561, "num_bytes" : 48138 } }, "config_name" : "fra-toki" }, "hbs-pol" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 33128, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 416, "dataset_name" : "tatoeba_mt", "num_bytes" : 33128, "name" : "test" } }, "config_name" : "hbs-pol", "dataset_size" : 416, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "epo-glg" : { "dataset_size" : 328, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 24476, "num_examples" : 328, "dataset_name" : "tatoeba_mt" } }, "config_name" : "epo-glg", "size_in_bytes" : 24476, "builder_name" : "tatoeba_mt" }, "spa-ukr" : { "dataset_size" : 23082, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 1869322, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 817523, "dataset_name" : "tatoeba_mt", "num_examples" : 10114 }, "validation" : { "num_examples" : 12968, "dataset_name" : "tatoeba_mt", "num_bytes" : 1051799, "name" : "validation" } }, "config_name" : "spa-ukr" }, "crh-tur" : { "config_name" : "crh-tur", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 207, "num_bytes" : 14520, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 14520, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 207 }, "nor-zho" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "nor-zho", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 200, "num_bytes" : 16243 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 16243, "supervised_keys" : null, "dataset_size" : 200, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "lfn_Cyrl-por" : { "config_name" : "lfn_Cyrl-por", "splits" : { "validation" : { "num_bytes" : 66291, "dataset_name" : "tatoeba_mt", "num_examples" : 445, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 219, "dataset_name" : "tatoeba_mt", "num_bytes" : 28489 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 94780, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 664 }, "eng-run" : { "config_name" : "eng-run", "splits" : { "test" : { "name" : "test", "num_bytes" : 90358, "dataset_name" : "tatoeba_mt", "num_examples" : 1702 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 90358, "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1702 }, "eng-pms" : { "size_in_bytes" : 21102, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 21102, "num_examples" : 268, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "eng-pms", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 268, "supervised_keys" : null }, "fra-lfn" : { "size_in_bytes" : 72065, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 55690, "dataset_name" : "tatoeba_mt", "num_examples" : 522 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 120, "num_bytes" : 16375, "name" : "validation" } }, "config_name" : "fra-lfn", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 642, "supervised_keys" : null }, "nld-nld" : { "config_name" : "nld-nld", "splits" : { "test" : { "num_bytes" : 74782, "dataset_name" : "tatoeba_mt", "num_examples" : 999, "name" : "test" }, "validation" : { "num_bytes" : 73790, "num_examples" : 1030, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 148572, "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 2029 }, "nld-por" : { "size_in_bytes" : 550628, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 185289, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 4881, "num_bytes" : 365339 } }, "config_name" : "nld-por", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 7380, "supervised_keys" : null }, "isl-ita" : { "supervised_keys" : null, "dataset_size" : 235, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 16140, "config_name" : "isl-ita", "splits" : { "test" : { "num_bytes" : 16140, "num_examples" : 235, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "aze-spa" : { "dataset_size" : 209, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 13994, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 13994, "dataset_name" : "tatoeba_mt", "num_examples" : 209, "name" : "test" } }, "config_name" : "aze-spa" }, "hun-tur" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 63469, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1001, "num_bytes" : 63469 } }, "config_name" : "hun-tur", "dataset_size" : 1001, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "epo-ita" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10302, "num_bytes" : 765268, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 40618, "dataset_name" : "tatoeba_mt", "num_bytes" : 2967019 } }, "config_name" : "epo-ita", "size_in_bytes" : 3732287, "builder_name" : "tatoeba_mt", "dataset_size" : 50920, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 } }, "ceb-eng" : { "config_name" : "ceb-eng", "splits" : { "test" : { "name" : "test", "num_bytes" : 23173, "num_examples" : 377, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 23173, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 377 }, "fra-heb" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 311865, "dataset_name" : "tatoeba_mt", "num_examples" : 3280 }, "validation" : { "num_examples" : 1063, "dataset_name" : "tatoeba_mt", "num_bytes" : 100195, "name" : "validation" } }, "config_name" : "fra-heb", "size_in_bytes" : 412060, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 4343, "supervised_keys" : null }, "eng-srp_Cyrl" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-srp_Cyrl", "splits" : { "test" : { "num_bytes" : 136541, "dataset_name" : "tatoeba_mt", "num_examples" : 1579, "name" : "test" }, "validation" : { "num_bytes" : 875779, "dataset_name" : "tatoeba_mt", "num_examples" : 8816, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1012320, "supervised_keys" : null, "dataset_size" : 10395, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "hrv-spa" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 253, "config_name" : "hrv-spa", "splits" : { "test" : { "num_bytes" : 16960, "num_examples" : 253, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 16960, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-zsm_Latn" : { "dataset_size" : 1379, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 130624, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 81188, "dataset_name" : "tatoeba_mt", "num_examples" : 844 }, "test" : { "name" : "test", "num_examples" : 535, "dataset_name" : "tatoeba_mt", "num_bytes" : 49436 } }, "config_name" : "eng-zsm_Latn" }, "rus-sah" : { "dataset_size" : 993, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 993, "num_bytes" : 117992, "name" : "test" } }, "config_name" : "rus-sah", "size_in_bytes" : 117992, "builder_name" : "tatoeba_mt" }, "hun-hun" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 84532, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 1063, "dataset_name" : "tatoeba_mt", "num_bytes" : 84532 } }, "config_name" : "hun-hun", "dataset_size" : 1063, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "eng-hye" : { "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-hye", "splits" : { "test" : { "name" : "test", "num_bytes" : 73369, "dataset_name" : "tatoeba_mt", "num_examples" : 1120 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 999, "num_bytes" : 66759, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 140128, "supervised_keys" : null, "dataset_size" : 2119, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "rus-cmn_Hans" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 116284, "dataset_name" : "tatoeba_mt", "num_examples" : 1085 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 2220, "num_bytes" : 239435 } }, "config_name" : "rus-cmn_Hans", "size_in_bytes" : 355719, "builder_name" : "tatoeba_mt", "dataset_size" : 3305, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "dan-fra" : { "supervised_keys" : null, "dataset_size" : 1774, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "dan-fra", "splits" : { "test" : { "num_bytes" : 120949, "num_examples" : 1730, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 4617, "dataset_name" : "tatoeba_mt", "num_examples" : 44 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 125566 }, "dsb-slv" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 16248, "config_name" : "dsb-slv", "splits" : { "test" : { "name" : "test", "num_bytes" : 16248, "dataset_name" : "tatoeba_mt", "num_examples" : 329 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 329 }, "deu-frr" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 277, "config_name" : "deu-frr", "splits" : { "test" : { "name" : "test", "num_examples" : 277, "dataset_name" : "tatoeba_mt", "num_bytes" : 19409 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 19409, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "rus-toki_Latn" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 898, "supervised_keys" : null, "size_in_bytes" : 86569, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 898, "num_bytes" : 86569, "name" : "test" } }, "config_name" : "rus-toki_Latn", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "grn-spa" : { "splits" : { "test" : { "num_bytes" : 45645, "num_examples" : 658, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 129, "dataset_name" : "tatoeba_mt", "num_bytes" : 11168 } }, "config_name" : "grn-spa", "size_in_bytes" : 56813, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 787, "supervised_keys" : null }, "deu-ind" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 42298, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 496, "num_bytes" : 42298, "name" : "test" } }, "config_name" : "deu-ind", "dataset_size" : 496, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "dan-dan" : { "dataset_size" : 211, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 211, "num_bytes" : 18075 } }, "config_name" : "dan-dan", "size_in_bytes" : 18075, "builder_name" : "tatoeba_mt" }, "eng-uzb" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 457, "config_name" : "eng-uzb", "splits" : { "test" : { "num_bytes" : 31244, "num_examples" : 457, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 31244, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ara-spa" : { "splits" : { "test" : { "name" : "test", "num_examples" : 1510, "dataset_name" : "tatoeba_mt", "num_bytes" : 118851 }, "validation" : { "name" : "validation", "num_bytes" : 81989, "num_examples" : 1030, "dataset_name" : "tatoeba_mt" } }, "config_name" : "ara-spa", "size_in_bytes" : 200840, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 2540, "supervised_keys" : null }, "eng-ron" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 15167, "supervised_keys" : null, "size_in_bytes" : 1207493, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 5507, "dataset_name" : "tatoeba_mt", "num_bytes" : 429293, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 9660, "num_bytes" : 778200 } }, "config_name" : "eng-ron", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "ron-rus" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 66481, "config_name" : "ron-rus", "splits" : { "test" : { "name" : "test", "num_bytes" : 66481, "dataset_name" : "tatoeba_mt", "num_examples" : 781 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 781 }, "eng-lfn_Cyrl" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 216014, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 1219, "dataset_name" : "tatoeba_mt", "num_bytes" : 136461, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 846, "dataset_name" : "tatoeba_mt", "num_bytes" : 79553 } }, "config_name" : "eng-lfn_Cyrl", "dataset_size" : 2065, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "nor-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 171086, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 171086, "dataset_name" : "tatoeba_mt", "num_examples" : 1278 } }, "config_name" : "nor-rus", "dataset_size" : 1278, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "epo-hbs" : { "supervised_keys" : null, "dataset_size" : 5300, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "epo-hbs", "splits" : { "test" : { "num_examples" : 2499, "dataset_name" : "tatoeba_mt", "num_bytes" : 258348, "name" : "test" }, "validation" : { "num_bytes" : 288686, "dataset_name" : "tatoeba_mt", "num_examples" : 2801, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 547034 }, "epo-jpn_Hani" : { "splits" : { "validation" : { "num_examples" : 657, "dataset_name" : "tatoeba_mt", "num_bytes" : 58227, "name" : "validation" }, "test" : { "num_examples" : 327, "dataset_name" : "tatoeba_mt", "num_bytes" : 31274, "name" : "test" } }, "config_name" : "epo-jpn_Hani", "size_in_bytes" : 89501, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 984, "supervised_keys" : null }, "tat-vie" : { "dataset_size" : 283, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 283, "num_bytes" : 30530, "name" : "test" } }, "config_name" : "tat-vie", "size_in_bytes" : 30530, "builder_name" : "tatoeba_mt" }, "ell-swe" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 252, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 252, "num_bytes" : 21696 } }, "config_name" : "ell-swe", "size_in_bytes" : 21696, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fra-toki_Latn" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 560, "builder_name" : "tatoeba_mt", "size_in_bytes" : 48117, "config_name" : "fra-toki_Latn", "splits" : { "test" : { "num_bytes" : 48117, "num_examples" : 560, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "cmn_Hans-wuu" : { "supervised_keys" : null, "dataset_size" : 2647, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "cmn_Hans-wuu", "splits" : { "test" : { "name" : "test", "num_bytes" : 73864, "dataset_name" : "tatoeba_mt", "num_examples" : 814 }, "validation" : { "name" : "validation", "num_bytes" : 162314, "dataset_name" : "tatoeba_mt", "num_examples" : 1833 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 236178 }, "ukr-ukr" : { "size_in_bytes" : 73415, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 73415, "dataset_name" : "tatoeba_mt", "num_examples" : 823, "name" : "test" } }, "config_name" : "ukr-ukr", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 823, "supervised_keys" : null }, "fra-cmn_Hans" : { "config_name" : "fra-cmn_Hans", "splits" : { "test" : { "name" : "test", "num_examples" : 6824, "dataset_name" : "tatoeba_mt", "num_bytes" : 632952 }, "validation" : { "num_examples" : 5146, "dataset_name" : "tatoeba_mt", "num_bytes" : 478104, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1111056, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 11970 }, "spa-yid" : { "dataset_size" : 448, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 42161, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_examples" : 42, "dataset_name" : "tatoeba_mt", "num_bytes" : 4666 }, "test" : { "name" : "test", "num_bytes" : 37495, "dataset_name" : "tatoeba_mt", "num_examples" : 406 } }, "config_name" : "spa-yid" }, "lit-spa" : { "dataset_size" : 453, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 29135, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 29135, "num_examples" : 453, "dataset_name" : "tatoeba_mt" } }, "config_name" : "lit-spa" }, "jpn_Kana-rus" : { "dataset_size" : 835, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "num_bytes" : 64911, "num_examples" : 536, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 299, "num_bytes" : 37932, "name" : "test" } }, "config_name" : "jpn_Kana-rus", "size_in_bytes" : 102843, "builder_name" : "tatoeba_mt" }, "dan-nor" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 182033, "config_name" : "dan-nor", "splits" : { "validation" : { "num_bytes" : 79691, "dataset_name" : "tatoeba_mt", "num_examples" : 1016, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 102342, "num_examples" : 1310, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2326 }, "hbs-ita" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 37083, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 37083, "dataset_name" : "tatoeba_mt", "num_examples" : 533 } }, "config_name" : "hbs-ita", "dataset_size" : 533, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "hun-swe" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2137, "builder_name" : "tatoeba_mt", "size_in_bytes" : 153769, "config_name" : "hun-swe", "splits" : { "test" : { "name" : "test", "num_bytes" : 117937, "num_examples" : 1613, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 35832, "num_examples" : 524, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ita-nor" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 938, "config_name" : "ita-nor", "splits" : { "test" : { "num_examples" : 938, "dataset_name" : "tatoeba_mt", "num_bytes" : 102588, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 102588, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "hun-kor_Hang" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 267, "supervised_keys" : null, "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 267, "num_bytes" : 22473, "name" : "test" } }, "config_name" : "hun-kor_Hang", "size_in_bytes" : 22473, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "rus-slv" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 700, "supervised_keys" : null, "size_in_bytes" : 61006, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 656, "num_bytes" : 56930, "name" : "test" }, "validation" : { "num_bytes" : 4076, "dataset_name" : "tatoeba_mt", "num_examples" : 44, "name" : "validation" } }, "config_name" : "rus-slv", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "bel-pol" : { "supervised_keys" : null, "dataset_size" : 286, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 26236, "config_name" : "bel-pol", "splits" : { "test" : { "num_bytes" : 26236, "num_examples" : 286, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "eng-lat" : { "supervised_keys" : null, "dataset_size" : 24332, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-lat", "splits" : { "test" : { "num_bytes" : 968004, "dataset_name" : "tatoeba_mt", "num_examples" : 10297, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 14035, "num_bytes" : 1325827 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2293831 }, "eng-kor" : { "dataset_size" : 3439, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 300058, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 92025, "num_examples" : 1040, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 208033, "dataset_name" : "tatoeba_mt", "num_examples" : 2399 } }, "config_name" : "eng-kor" }, "ces-slv" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 488, "supervised_keys" : null, "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 488, "num_bytes" : 28804, "name" : "test" } }, "config_name" : "ces-slv", "size_in_bytes" : 28804, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-tuk_Latn" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 2498, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_bytes" : 213379, "dataset_name" : "tatoeba_mt", "num_examples" : 2498 } }, "config_name" : "eng-tuk_Latn", "size_in_bytes" : 213379, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "hbs-spa" : { "size_in_bytes" : 42872, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 42872, "dataset_name" : "tatoeba_mt", "num_examples" : 606, "name" : "test" } }, "config_name" : "hbs-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 606, "supervised_keys" : null }, "hin-zho" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 44648, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 44648, "dataset_name" : "tatoeba_mt", "num_examples" : 323 } }, "config_name" : "hin-zho", "dataset_size" : 323, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "eng-tlh" : { "dataset_size" : 13418, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "num_bytes" : 540841, "dataset_name" : "tatoeba_mt", "num_examples" : 8419, "name" : "validation" }, "test" : { "num_bytes" : 321225, "num_examples" : 4999, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "eng-tlh", "size_in_bytes" : 862066, "builder_name" : "tatoeba_mt" }, "jpn-pol" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 2598341, "config_name" : "jpn-pol", "splits" : { "test" : { "num_examples" : 9998, "dataset_name" : "tatoeba_mt", "num_bytes" : 1044875, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 14808, "num_bytes" : 1553466, "name" : "validation" } }, "supervised_keys" : null, "dataset_size" : 24806, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-jpn_Hira" : { "dataset_size" : 182676, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 19971469, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_examples" : 170846, "dataset_name" : "tatoeba_mt", "num_bytes" : 18681476 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 11830, "num_bytes" : 1289993 } }, "config_name" : "eng-jpn_Hira" }, "heb-pol" : { "dataset_size" : 12755, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 1119495, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_examples" : 7756, "dataset_name" : "tatoeba_mt", "num_bytes" : 682085 }, "test" : { "num_bytes" : 437410, "num_examples" : 4999, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "heb-pol" }, "spa-zho" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 9625, "config_name" : "spa-zho", "splits" : { "test" : { "name" : "test", "num_bytes" : 195859, "num_examples" : 2704, "dataset_name" : "tatoeba_mt" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 6921, "num_bytes" : 509628 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 705487, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-gos" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 56438, "dataset_name" : "tatoeba_mt", "num_examples" : 1153 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 95, "num_bytes" : 6300 } }, "config_name" : "eng-gos", "size_in_bytes" : 62738, "builder_name" : "tatoeba_mt", "dataset_size" : 1248, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "nor-ukr" : { "size_in_bytes" : 59241, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 59241, "num_examples" : 669, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "nor-ukr", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 669, "supervised_keys" : null }, "fra-slv" : { "size_in_bytes" : 38259, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 38259, "dataset_name" : "tatoeba_mt", "num_examples" : 447, "name" : "test" } }, "config_name" : "fra-slv", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 447, "supervised_keys" : null }, "ara-ita" : { "supervised_keys" : null, "dataset_size" : 234, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 18780, "config_name" : "ara-ita", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 234, "num_bytes" : 18780 } } }, "deu-swg" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 166882, "num_examples" : 1522, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 16228, "dataset_name" : "tatoeba_mt", "num_examples" : 161, "name" : "validation" } }, "config_name" : "deu-swg", "size_in_bytes" : 183110, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 1683, "supervised_keys" : null }, "epo-tgl" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 101982, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 101982, "num_examples" : 1108, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "epo-tgl", "dataset_size" : 1108, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "aze_Latn-tur" : { "supervised_keys" : null, "dataset_size" : 1140, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 80884, "config_name" : "aze_Latn-tur", "splits" : { "test" : { "num_bytes" : 80884, "dataset_name" : "tatoeba_mt", "num_examples" : 1140, "name" : "test" } } }, "deu-eng" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 26733054, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 1595470, "dataset_name" : "tatoeba_mt", "num_examples" : 17564, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 289778, "dataset_name" : "tatoeba_mt", "num_bytes" : 25137584 } }, "config_name" : "deu-eng", "dataset_size" : 307342, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "deu-ina" : { "size_in_bytes" : 145336, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 114, "num_bytes" : 8179 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1256, "num_bytes" : 137157 } }, "config_name" : "deu-ina", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 1370, "supervised_keys" : null }, "isl-spa" : { "size_in_bytes" : 13806, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 237, "num_bytes" : 13806 } }, "config_name" : "isl-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 237, "supervised_keys" : null }, "ita-pol" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 2294, "supervised_keys" : null, "size_in_bytes" : 167582, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_examples" : 1001, "dataset_name" : "tatoeba_mt", "num_bytes" : 73569 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1293, "num_bytes" : 94013 } }, "config_name" : "ita-pol", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "ukr-zho" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 138693, "config_name" : "ukr-zho", "splits" : { "test" : { "name" : "test", "num_bytes" : 138693, "num_examples" : 1574, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1574 }, "epo-spa" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "epo-spa", "splits" : { "validation" : { "num_bytes" : 5443363, "dataset_name" : "tatoeba_mt", "num_examples" : 65095, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 10734, "dataset_name" : "tatoeba_mt", "num_bytes" : 894253 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 6337616, "supervised_keys" : null, "dataset_size" : 75829, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-ido_Latn" : { "supervised_keys" : null, "dataset_size" : 1958, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-ido_Latn", "splits" : { "test" : { "name" : "test", "num_bytes" : 149043, "num_examples" : 1958, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 149043 }, "ell-tur" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 9099, "num_examples" : 108, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_examples" : 1468, "dataset_name" : "tatoeba_mt", "num_bytes" : 132670 } }, "config_name" : "ell-tur", "size_in_bytes" : 141769, "builder_name" : "tatoeba_mt", "dataset_size" : 1576, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "jpn_Hani-por" : { "dataset_size" : 494, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 159, "num_bytes" : 14546 }, "test" : { "name" : "test", "num_bytes" : 30661, "num_examples" : 335, "dataset_name" : "tatoeba_mt" } }, "config_name" : "jpn_Hani-por", "size_in_bytes" : 45207, "builder_name" : "tatoeba_mt" }, "eng-ota" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 687, "config_name" : "eng-ota", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 687, "num_bytes" : 58713 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 58713, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-kaz" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 37502, "config_name" : "eng-kaz", "splits" : { "test" : { "name" : "test", "num_bytes" : 37502, "dataset_name" : "tatoeba_mt", "num_examples" : 402 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 402 }, "bul-fra" : { "supervised_keys" : null, "dataset_size" : 445, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "bul-fra", "splits" : { "test" : { "num_bytes" : 48703, "dataset_name" : "tatoeba_mt", "num_examples" : 445, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 48703 }, "eng-kur" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 36707, "config_name" : "eng-kur", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 79, "num_bytes" : 5812 }, "test" : { "num_bytes" : 30895, "dataset_name" : "tatoeba_mt", "num_examples" : 441, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 520 }, "ina-tlh" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 248, "builder_name" : "tatoeba_mt", "size_in_bytes" : 14415, "config_name" : "ina-tlh", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 215, "num_bytes" : 12352 }, "validation" : { "name" : "validation", "num_examples" : 33, "dataset_name" : "tatoeba_mt", "num_bytes" : 2063 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "pol-spa" : { "size_in_bytes" : 589387, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 5074, "num_bytes" : 394091 }, "test" : { "num_examples" : 2543, "dataset_name" : "tatoeba_mt", "num_bytes" : 195296, "name" : "test" } }, "config_name" : "pol-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 7617, "supervised_keys" : null }, "eng-ile" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 185479, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 72927, "dataset_name" : "tatoeba_mt", "num_examples" : 1107, "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1710, "num_bytes" : 112552, "name" : "test" } }, "config_name" : "eng-ile", "dataset_size" : 2817, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "eus-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 35444, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 35444, "dataset_name" : "tatoeba_mt", "num_examples" : 484 } }, "config_name" : "eus-rus", "dataset_size" : 484, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "deu-lit" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 117833, "config_name" : "deu-lit", "splits" : { "test" : { "name" : "test", "num_examples" : 1114, "dataset_name" : "tatoeba_mt", "num_bytes" : 92530 }, "validation" : { "num_bytes" : 25303, "num_examples" : 354, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1468 }, "fra-por" : { "supervised_keys" : null, "dataset_size" : 27578, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fra-por", "splits" : { "test" : { "num_examples" : 10517, "dataset_name" : "tatoeba_mt", "num_bytes" : 858332, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 1353409, "dataset_name" : "tatoeba_mt", "num_examples" : 17061 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2211741 }, "jpn-nor" : { "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1046, "config_name" : "jpn-nor", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1046, "num_bytes" : 104634, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 104634, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "bua_Cyrl-rus" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 804, "config_name" : "bua_Cyrl-rus", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 804, "num_bytes" : 74416, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 74416, "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "hun-jpn" : { "dataset_size" : 6157, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_examples" : 3659, "dataset_name" : "tatoeba_mt", "num_bytes" : 409758 }, "test" : { "num_bytes" : 282076, "dataset_name" : "tatoeba_mt", "num_examples" : 2498, "name" : "test" } }, "config_name" : "hun-jpn", "size_in_bytes" : 691834, "builder_name" : "tatoeba_mt" }, "fra-hrv" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 19608, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 19608, "dataset_name" : "tatoeba_mt", "num_examples" : 257, "name" : "test" } }, "config_name" : "fra-hrv", "dataset_size" : 257, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "epo-oci" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 54830, "num_examples" : 587, "dataset_name" : "tatoeba_mt" } }, "config_name" : "epo-oci", "size_in_bytes" : 54830, "builder_name" : "tatoeba_mt", "dataset_size" : 587, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "fra-nld" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 28281, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 16734, "num_bytes" : 1233531 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 11547, "num_bytes" : 894048 } }, "config_name" : "fra-nld", "size_in_bytes" : 2127579, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "bel-fra" : { "supervised_keys" : null, "dataset_size" : 282, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "bel-fra", "splits" : { "test" : { "name" : "test", "num_examples" : 282, "dataset_name" : "tatoeba_mt", "num_bytes" : 26819 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 26819 }, "eng-uig" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-uig", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 3023, "num_bytes" : 314767, "name" : "test" }, "validation" : { "num_bytes" : 105302, "dataset_name" : "tatoeba_mt", "num_examples" : 1005, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 420069, "supervised_keys" : null, "dataset_size" : 4028, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "ina-lat" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 980, "supervised_keys" : null, "size_in_bytes" : 77145, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 4698, "num_examples" : 69, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 72447, "num_examples" : 911, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "ina-lat", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "ara-deu" : { "supervised_keys" : null, "dataset_size" : 2236, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 190274, "config_name" : "ara-deu", "splits" : { "test" : { "num_bytes" : 102672, "dataset_name" : "tatoeba_mt", "num_examples" : 1208, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 87602, "dataset_name" : "tatoeba_mt", "num_examples" : 1028 } } }, "vie-cmn_Hans" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 344, "builder_name" : "tatoeba_mt", "size_in_bytes" : 38707, "config_name" : "vie-cmn_Hans", "splits" : { "test" : { "num_bytes" : 38707, "num_examples" : 344, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "glg-spa" : { "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "glg-spa", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 86358, "num_examples" : 1011, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_examples" : 2120, "dataset_name" : "tatoeba_mt", "num_bytes" : 183901 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 270259, "supervised_keys" : null, "dataset_size" : 3131, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "bel-ita" : { "size_in_bytes" : 22431, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 22431, "num_examples" : 263, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "bel-ita", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 263, "supervised_keys" : null }, "dtp-eng" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 218377, "config_name" : "dtp-eng", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1926, "num_bytes" : 141337, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 77040, "num_examples" : 1011, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2937 }, "eng-mar" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 1139103, "num_examples" : 10395, "dataset_name" : "tatoeba_mt" }, "validation" : { "name" : "validation", "num_examples" : 43057, "dataset_name" : "tatoeba_mt", "num_bytes" : 4688040 } }, "config_name" : "eng-mar", "size_in_bytes" : 5827143, "builder_name" : "tatoeba_mt", "dataset_size" : 53452, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "eng-gle" : { "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-gle", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 26, "num_bytes" : 2051 }, "test" : { "name" : "test", "num_examples" : 1912, "dataset_name" : "tatoeba_mt", "num_bytes" : 118507 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 120558, "supervised_keys" : null, "dataset_size" : 1938, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "zho-zho" : { "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 5809, "num_bytes" : 502917, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 264336, "num_examples" : 3113, "dataset_name" : "tatoeba_mt" } }, "config_name" : "zho-zho", "size_in_bytes" : 767253, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 8922, "supervised_keys" : null }, "eng-mon" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 436, "supervised_keys" : null, "splits" : { "validation" : { "num_examples" : 22, "dataset_name" : "tatoeba_mt", "num_bytes" : 2922, "name" : "validation" }, "test" : { "num_bytes" : 46279, "dataset_name" : "tatoeba_mt", "num_examples" : 414, "name" : "test" } }, "config_name" : "eng-mon", "size_in_bytes" : 49201, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn_Hani-cmn_Hans" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 443, "builder_name" : "tatoeba_mt", "size_in_bytes" : 38519, "config_name" : "jpn_Hani-cmn_Hans", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 19663, "dataset_name" : "tatoeba_mt", "num_examples" : 228 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 215, "num_bytes" : 18856, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "lit-pol" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 126233, "dataset_name" : "tatoeba_mt", "num_examples" : 1786 } }, "config_name" : "lit-pol", "size_in_bytes" : 126233, "builder_name" : "tatoeba_mt", "dataset_size" : 1786, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "eng-sqi" : { "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1108, "num_bytes" : 85651 } }, "config_name" : "eng-sqi", "size_in_bytes" : 85651, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 1108, "supervised_keys" : null }, "nds-spa" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 922, "builder_name" : "tatoeba_mt", "size_in_bytes" : 59877, "config_name" : "nds-spa", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 922, "num_bytes" : 59877 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "nld-tur" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 470806, "config_name" : "nld-tur", "splits" : { "validation" : { "num_examples" : 3879, "dataset_name" : "tatoeba_mt", "num_bytes" : 286529, "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 184277, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 6378, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "fra-jpn" : { "config_name" : "fra-jpn", "splits" : { "test" : { "num_bytes" : 1102814, "num_examples" : 10168, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "num_examples" : 28593, "dataset_name" : "tatoeba_mt", "num_bytes" : 3095807, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 4198621, "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 38761 }, "hun-nld" : { "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2066, "num_examples" : 34, "dataset_name" : "tatoeba_mt" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1628, "num_bytes" : 113849, "name" : "test" } }, "config_name" : "hun-nld", "size_in_bytes" : 115915, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 1662, "supervised_keys" : null }, "ina-yid" : { "dataset_size" : 906, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 85872, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 12363, "num_examples" : 119, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 73509, "dataset_name" : "tatoeba_mt", "num_examples" : 787 } }, "config_name" : "ina-yid" }, "heb-ita" : { "supervised_keys" : null, "dataset_size" : 1734, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 145186, "config_name" : "heb-ita", "splits" : { "validation" : { "name" : "validation", "num_examples" : 29, "dataset_name" : "tatoeba_mt", "num_bytes" : 2202 }, "test" : { "name" : "test", "num_bytes" : 142984, "num_examples" : 1705, "dataset_name" : "tatoeba_mt" } } }, "eng-ukr" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 1165759, "num_examples" : 13126, "dataset_name" : "tatoeba_mt" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 159486, "num_bytes" : 13818313, "name" : "validation" } }, "config_name" : "eng-ukr", "size_in_bytes" : 14984072, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 172612, "supervised_keys" : null }, "hun-por" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 424688, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 176114, "num_examples" : 2499, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "num_bytes" : 248574, "num_examples" : 3516, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "hun-por", "dataset_size" : 6015, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "run-rus" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 86691, "dataset_name" : "tatoeba_mt", "num_examples" : 1250 } }, "config_name" : "run-rus", "size_in_bytes" : 86691, "builder_name" : "tatoeba_mt", "dataset_size" : 1250, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "eng-fry" : { "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 219, "config_name" : "eng-fry", "splits" : { "test" : { "num_examples" : 219, "dataset_name" : "tatoeba_mt", "num_bytes" : 15737, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 15737, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "nld-toki" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 43655, "config_name" : "nld-toki", "splits" : { "test" : { "num_bytes" : 43655, "dataset_name" : "tatoeba_mt", "num_examples" : 667, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 667, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "ido_Latn-lfn_Latn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 22663, "config_name" : "ido_Latn-lfn_Latn", "splits" : { "test" : { "num_bytes" : 20200, "dataset_name" : "tatoeba_mt", "num_examples" : 308, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 2463, "dataset_name" : "tatoeba_mt", "num_examples" : 37 } }, "supervised_keys" : null, "dataset_size" : 345, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-epo" : { "dataset_size" : 259348, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 20969382, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 19513865, "dataset_name" : "tatoeba_mt", "num_examples" : 241674, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 17674, "num_bytes" : 1455517 } }, "config_name" : "eng-epo" }, "ara-pol" : { "dataset_size" : 206, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 206, "num_bytes" : 16218 } }, "config_name" : "ara-pol", "size_in_bytes" : 16218, "builder_name" : "tatoeba_mt" }, "fra-ido" : { "size_in_bytes" : 22474, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 295, "dataset_name" : "tatoeba_mt", "num_bytes" : 20748, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 25, "num_bytes" : 1726, "name" : "validation" } }, "config_name" : "fra-ido", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 320, "supervised_keys" : null }, "bul-ita" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 630560, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 4005, "num_bytes" : 389298 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 241262 } }, "config_name" : "bul-ita", "dataset_size" : 6504, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 } }, "eng-ltz" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 18648, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 292, "num_bytes" : 18648 } }, "config_name" : "eng-ltz", "dataset_size" : 292, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "ita-jpn_Hani" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 391, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_bytes" : 27233, "num_examples" : 287, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 10247, "num_examples" : 104, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "ita-jpn_Hani", "size_in_bytes" : 37480, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fra-swe" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 2532, "config_name" : "fra-swe", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1126, "num_bytes" : 79547, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 1406, "dataset_name" : "tatoeba_mt", "num_bytes" : 101365 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 180912, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-isl" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 789981, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 6937, "dataset_name" : "tatoeba_mt", "num_bytes" : 579029, "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2502, "num_bytes" : 210952, "name" : "test" } }, "config_name" : "eng-isl", "dataset_size" : 9439, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 } }, "eng-nno" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 963, "config_name" : "eng-nno", "splits" : { "test" : { "num_bytes" : 34843, "dataset_name" : "tatoeba_mt", "num_examples" : 459, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 504, "num_bytes" : 38135, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 72978, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "por-cmn_Hant" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 385, "builder_name" : "tatoeba_mt", "size_in_bytes" : 26065, "config_name" : "por-cmn_Hant", "splits" : { "test" : { "num_bytes" : 26065, "dataset_name" : "tatoeba_mt", "num_examples" : 385, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ita-ita" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1348, "builder_name" : "tatoeba_mt", "size_in_bytes" : 107818, "config_name" : "ita-ita", "splits" : { "validation" : { "num_bytes" : 27857, "dataset_name" : "tatoeba_mt", "num_examples" : 349, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 999, "dataset_name" : "tatoeba_mt", "num_bytes" : 79961 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "lfn-rus" : { "dataset_size" : 249, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 28988, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 25529, "dataset_name" : "tatoeba_mt", "num_examples" : 224, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 25, "num_bytes" : 3459 } }, "config_name" : "lfn-rus" }, "deu-vie" : { "dataset_size" : 399, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 38820, "num_examples" : 399, "dataset_name" : "tatoeba_mt" } }, "config_name" : "deu-vie", "size_in_bytes" : 38820, "builder_name" : "tatoeba_mt" }, "fra-kor_Hang" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 31711, "config_name" : "fra-kor_Hang", "splits" : { "test" : { "name" : "test", "num_bytes" : 31711, "num_examples" : 315, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 315 }, "ber_Latn-fra" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 3345793, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 34697, "num_bytes" : 2588010 }, "test" : { "name" : "test", "num_bytes" : 757783, "dataset_name" : "tatoeba_mt", "num_examples" : 10139 } }, "config_name" : "ber_Latn-fra", "dataset_size" : 44836, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "bel-deu" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 56902, "config_name" : "bel-deu", "splits" : { "test" : { "name" : "test", "num_bytes" : 56902, "num_examples" : 550, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 550, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-jbo" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 11934, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 4995, "num_bytes" : 364043 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 6939, "num_bytes" : 510193 } }, "config_name" : "eng-jbo", "size_in_bytes" : 874236, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-nov" : { "supervised_keys" : null, "dataset_size" : 216, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 14871, "config_name" : "eng-nov", "splits" : { "test" : { "name" : "test", "num_examples" : 216, "dataset_name" : "tatoeba_mt", "num_bytes" : 14871 } } }, "ara-fra" : { "config_name" : "ara-fra", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 3669, "dataset_name" : "tatoeba_mt", "num_examples" : 35 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1568, "num_bytes" : 134171 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 137840, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1603 }, "jpn_Hira-rus" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 22793, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2835585, "config_name" : "jpn_Hira-rus", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 9007, "num_bytes" : 1119177, "name" : "test" }, "validation" : { "num_bytes" : 1716408, "dataset_name" : "tatoeba_mt", "num_examples" : 13786, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "dan-spa" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 10134, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "num_bytes" : 389803, "dataset_name" : "tatoeba_mt", "num_examples" : 5135 }, "test" : { "num_bytes" : 378035, "num_examples" : 4999, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "dan-spa", "size_in_bytes" : 767838, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ber-eng" : { "dataset_size" : 119384, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 9791456, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 8894631, "dataset_name" : "tatoeba_mt", "num_examples" : 108421, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 896825, "num_examples" : 10963, "dataset_name" : "tatoeba_mt" } }, "config_name" : "ber-eng" }, "eng-yid" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 4373, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1891, "num_bytes" : 180143 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2482, "num_bytes" : 232158, "name" : "test" } }, "config_name" : "eng-yid", "size_in_bytes" : 412301, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "hrv-pol" : { "dataset_size" : 270, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_examples" : 270, "dataset_name" : "tatoeba_mt", "num_bytes" : 19473 } }, "config_name" : "hrv-pol", "size_in_bytes" : 19473, "builder_name" : "tatoeba_mt" }, "fra-tlh_Latn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 41512, "config_name" : "fra-tlh_Latn", "splits" : { "test" : { "name" : "test", "num_bytes" : 41512, "dataset_name" : "tatoeba_mt", "num_examples" : 618 } }, "supervised_keys" : null, "dataset_size" : 618, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-jpn_Kana" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 3730, "supervised_keys" : null, "size_in_bytes" : 396514, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 502, "dataset_name" : "tatoeba_mt", "num_bytes" : 51632, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 3228, "num_bytes" : 344882, "name" : "validation" } }, "config_name" : "eng-jpn_Kana", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "jpn_Hira-jpn_Hira" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 440, "config_name" : "jpn_Hira-jpn_Hira", "splits" : { "test" : { "name" : "test", "num_bytes" : 60247, "dataset_name" : "tatoeba_mt", "num_examples" : 440 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 60247, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "rus-swe" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 1332, "supervised_keys" : null, "size_in_bytes" : 121345, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 1281, "dataset_name" : "tatoeba_mt", "num_bytes" : 116623, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 51, "num_bytes" : 4722, "name" : "validation" } }, "config_name" : "rus-swe", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "bul-deu" : { "dataset_size" : 313, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 30294, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 313, "num_bytes" : 30294, "name" : "test" } }, "config_name" : "bul-deu" }, "ita-swe" : { "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 714, "builder_name" : "tatoeba_mt", "size_in_bytes" : 51351, "config_name" : "ita-swe", "splits" : { "test" : { "num_bytes" : 51351, "num_examples" : 714, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "lad-lat" : { "splits" : { "validation" : { "num_bytes" : 2277, "num_examples" : 32, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 213, "num_bytes" : 14669, "name" : "test" } }, "config_name" : "lad-lat", "size_in_bytes" : 16946, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 245, "supervised_keys" : null }, "por-ron" : { "dataset_size" : 710, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 51126, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 30, "num_bytes" : 1971 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 680, "num_bytes" : 49155, "name" : "test" } }, "config_name" : "por-ron" }, "jpn-jpn" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 74694, "config_name" : "jpn-jpn", "splits" : { "test" : { "num_examples" : 583, "dataset_name" : "tatoeba_mt", "num_bytes" : 74694, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 583 }, "fra-ita" : { "supervised_keys" : null, "dataset_size" : 78957, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 5612931, "config_name" : "fra-ita", "splits" : { "test" : { "num_bytes" : 718494, "dataset_name" : "tatoeba_mt", "num_examples" : 10090, "name" : "test" }, "validation" : { "num_bytes" : 4894437, "num_examples" : 68867, "dataset_name" : "tatoeba_mt", "name" : "validation" } } }, "eng-ina" : { "config_name" : "eng-ina", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 574881, "dataset_name" : "tatoeba_mt", "num_examples" : 6233 }, "test" : { "name" : "test", "num_bytes" : 473922, "dataset_name" : "tatoeba_mt", "num_examples" : 4987 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1048803, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 11220 }, "eng-eng" : { "dataset_size" : 108668, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_examples" : 96607, "dataset_name" : "tatoeba_mt", "num_bytes" : 9036672 }, "test" : { "num_bytes" : 1124615, "dataset_name" : "tatoeba_mt", "num_examples" : 12061, "name" : "test" } }, "config_name" : "eng-eng", "size_in_bytes" : 10161287, "builder_name" : "tatoeba_mt" }, "cor-fra" : { "size_in_bytes" : 31711, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 554, "dataset_name" : "tatoeba_mt", "num_bytes" : 31711 } }, "config_name" : "cor-fra", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 554, "supervised_keys" : null }, "heb-jpn" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 240, "config_name" : "heb-jpn", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 240, "num_bytes" : 27471, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 27471, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fin-zho" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 381, "config_name" : "fin-zho", "splits" : { "test" : { "num_examples" : 381, "dataset_name" : "tatoeba_mt", "num_bytes" : 24985, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 24985, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "epo-heb" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 29973, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2868964, "config_name" : "epo-heb", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 19606, "num_bytes" : 1880897, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 10367, "num_bytes" : 988067 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "deu-kur" : { "supervised_keys" : null, "dataset_size" : 236, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 16071, "config_name" : "deu-kur", "splits" : { "test" : { "name" : "test", "num_bytes" : 16071, "dataset_name" : "tatoeba_mt", "num_examples" : 236 } } }, "jbo-zho" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 539, "builder_name" : "tatoeba_mt", "size_in_bytes" : 38054, "config_name" : "jbo-zho", "splits" : { "test" : { "num_bytes" : 36339, "num_examples" : 517, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 1715, "dataset_name" : "tatoeba_mt", "num_examples" : 22 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "srp_Cyrl-rus" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 2289, "supervised_keys" : null, "size_in_bytes" : 265320, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 166638, "num_examples" : 1409, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 98682, "num_examples" : 880, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "srp_Cyrl-rus", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "bul-jpn" : { "dataset_size" : 320, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 39835, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 39835, "dataset_name" : "tatoeba_mt", "num_examples" : 320, "name" : "test" } }, "config_name" : "bul-jpn" }, "lat-tlh" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 261, "supervised_keys" : null, "size_in_bytes" : 15864, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 232, "dataset_name" : "tatoeba_mt", "num_bytes" : 14111, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 1753, "dataset_name" : "tatoeba_mt", "num_examples" : 29 } }, "config_name" : "lat-tlh", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "epo-lad" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 416, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_bytes" : 25590, "num_examples" : 365, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 3567, "dataset_name" : "tatoeba_mt", "num_examples" : 51, "name" : "validation" } }, "config_name" : "epo-lad", "size_in_bytes" : 29157, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "nld-spa" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 2266943, "config_name" : "nld-spa", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 1450714, "dataset_name" : "tatoeba_mt", "num_examples" : 18067 }, "test" : { "name" : "test", "num_examples" : 10112, "dataset_name" : "tatoeba_mt", "num_bytes" : 816229 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 28179 }, "eng-lit" : { "dataset_size" : 8169, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 580668, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 400501, "dataset_name" : "tatoeba_mt", "num_examples" : 5642, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2527, "num_bytes" : 180167 } }, "config_name" : "eng-lit" }, "ces-ita" : { "supervised_keys" : null, "dataset_size" : 1098, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 85180, "config_name" : "ces-ita", "splits" : { "test" : { "name" : "test", "num_bytes" : 85180, "dataset_name" : "tatoeba_mt", "num_examples" : 1098 } } }, "deu-ile" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 3373, "config_name" : "deu-ile", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 91466, "dataset_name" : "tatoeba_mt", "num_examples" : 1371 }, "test" : { "name" : "test", "num_bytes" : 137021, "num_examples" : 2002, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 228487, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "afr-rus" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 227, "supervised_keys" : null, "size_in_bytes" : 17527, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 17527, "num_examples" : 227, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "afr-rus", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } } }, "spa-cmn_Hant" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 5262, "supervised_keys" : null, "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 3792, "num_bytes" : 278564, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 105555, "num_examples" : 1470, "dataset_name" : "tatoeba_mt" } }, "config_name" : "spa-cmn_Hant", "size_in_bytes" : 384119, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-rom" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 48250, "dataset_name" : "tatoeba_mt", "num_examples" : 705 } }, "config_name" : "eng-rom", "size_in_bytes" : 48250, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 705, "supervised_keys" : null }, "lat-lat" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 234, "builder_name" : "tatoeba_mt", "size_in_bytes" : 20141, "config_name" : "lat-lat", "splits" : { "test" : { "num_bytes" : 20141, "num_examples" : 234, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "epo-lfn" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 1258, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "num_bytes" : 22254, "dataset_name" : "tatoeba_mt", "num_examples" : 262 }, "test" : { "num_examples" : 996, "dataset_name" : "tatoeba_mt", "num_bytes" : 73629, "name" : "test" } }, "config_name" : "epo-lfn", "size_in_bytes" : 95883, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "tur-tur" : { "supervised_keys" : null, "dataset_size" : 3685, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "tur-tur", "splits" : { "validation" : { "name" : "validation", "num_examples" : 1186, "dataset_name" : "tatoeba_mt", "num_bytes" : 86899 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 182115, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 269014 }, "deu-lat" : { "splits" : { "validation" : { "name" : "validation", "num_examples" : 1101, "dataset_name" : "tatoeba_mt", "num_bytes" : 81264 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2015, "num_bytes" : 142634 } }, "config_name" : "deu-lat", "size_in_bytes" : 223898, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 3116, "supervised_keys" : null }, "fin-rus" : { "dataset_size" : 4650, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 91016, "dataset_name" : "tatoeba_mt", "num_examples" : 1008 }, "test" : { "num_bytes" : 323616, "num_examples" : 3642, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "fin-rus", "size_in_bytes" : 414632, "builder_name" : "tatoeba_mt" }, "uig_Arab-cmn_Hant" : { "config_name" : "uig_Arab-cmn_Hant", "splits" : { "test" : { "num_bytes" : 55221, "dataset_name" : "tatoeba_mt", "num_examples" : 541, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 55221, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 541 }, "hun-pol" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 140082, "dataset_name" : "tatoeba_mt", "num_examples" : 1933 } }, "config_name" : "hun-pol", "size_in_bytes" : 140082, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 1933, "supervised_keys" : null }, "ukr-cmn_Hant" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 42184, "dataset_name" : "tatoeba_mt", "num_examples" : 529 } }, "config_name" : "ukr-cmn_Hant", "size_in_bytes" : 42184, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 529, "supervised_keys" : null }, "epo-cmn_Hans" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 2010, "builder_name" : "tatoeba_mt", "size_in_bytes" : 159586, "config_name" : "epo-cmn_Hans", "splits" : { "validation" : { "name" : "validation", "num_examples" : 607, "dataset_name" : "tatoeba_mt", "num_bytes" : 48203 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1403, "num_bytes" : 111383 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "dan-tur" : { "config_name" : "dan-tur", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 757, "num_bytes" : 44770, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 44770, "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 757 }, "grn-por" : { "supervised_keys" : null, "dataset_size" : 296, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 23447, "config_name" : "grn-por", "splits" : { "test" : { "num_bytes" : 17641, "num_examples" : 223, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "num_examples" : 73, "dataset_name" : "tatoeba_mt", "num_bytes" : 5806, "name" : "validation" } } }, "nob-swe" : { "supervised_keys" : null, "dataset_size" : 562, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 38582, "config_name" : "nob-swe", "splits" : { "test" : { "num_examples" : 562, "dataset_name" : "tatoeba_mt", "num_bytes" : 38582, "name" : "test" } } }, "jbo-rus" : { "splits" : { "test" : { "name" : "test", "num_examples" : 1198, "dataset_name" : "tatoeba_mt", "num_bytes" : 109668 } }, "config_name" : "jbo-rus", "size_in_bytes" : 109668, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 1198, "supervised_keys" : null }, "chv-tur" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 44708, "config_name" : "chv-tur", "splits" : { "test" : { "name" : "test", "num_examples" : 507, "dataset_name" : "tatoeba_mt", "num_bytes" : 44708 } }, "supervised_keys" : null, "dataset_size" : 507, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "swe-swe" : { "size_in_bytes" : 72315, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1021, "num_bytes" : 72315, "name" : "test" } }, "config_name" : "swe-swe", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 1021, "supervised_keys" : null }, "deu-kor" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1103, "num_bytes" : 110132, "name" : "test" } }, "config_name" : "deu-kor", "size_in_bytes" : 110132, "builder_name" : "tatoeba_mt", "dataset_size" : 1103, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 } }, "deu-gos" : { "splits" : { "test" : { "name" : "test", "num_examples" : 206, "dataset_name" : "tatoeba_mt", "num_bytes" : 12260 } }, "config_name" : "deu-gos", "size_in_bytes" : 12260, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 206, "supervised_keys" : null }, "kor_Hang-rus" : { "splits" : { "test" : { "name" : "test", "num_examples" : 219, "dataset_name" : "tatoeba_mt", "num_bytes" : 23000 } }, "config_name" : "kor_Hang-rus", "size_in_bytes" : 23000, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 219, "supervised_keys" : null }, "deu-jpn_Hira" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2787804, "num_examples" : 25378, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 1070014, "num_examples" : 9809, "dataset_name" : "tatoeba_mt" } }, "config_name" : "deu-jpn_Hira", "size_in_bytes" : 3857818, "builder_name" : "tatoeba_mt", "dataset_size" : 35187, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "ita-jpn" : { "supervised_keys" : null, "dataset_size" : 3663, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "ita-jpn", "splits" : { "validation" : { "num_examples" : 1004, "dataset_name" : "tatoeba_mt", "num_bytes" : 106080, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 2659, "dataset_name" : "tatoeba_mt", "num_bytes" : 273616 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 379696 }, "ber-epo" : { "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "ber-epo", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1575, "num_bytes" : 133156 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 133156, "supervised_keys" : null, "dataset_size" : 1575, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "jpn-swe" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 895, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_bytes" : 82973, "dataset_name" : "tatoeba_mt", "num_examples" : 895 } }, "config_name" : "jpn-swe", "size_in_bytes" : 82973, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn_Hani-nld" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 543, "supervised_keys" : null, "size_in_bytes" : 44816, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 9513, "num_examples" : 118, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 35303, "dataset_name" : "tatoeba_mt", "num_examples" : 425, "name" : "test" } }, "config_name" : "jpn_Hani-nld", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } } }, "deu-tlh" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2132, "config_name" : "deu-tlh", "splits" : { "validation" : { "num_bytes" : 72280, "dataset_name" : "tatoeba_mt", "num_examples" : 1034, "name" : "validation" }, "test" : { "num_bytes" : 78188, "dataset_name" : "tatoeba_mt", "num_examples" : 1098, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 150468, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ces-deu" : { "splits" : { "test" : { "name" : "test", "num_examples" : 3489, "dataset_name" : "tatoeba_mt", "num_bytes" : 282537 }, "validation" : { "num_examples" : 1126, "dataset_name" : "tatoeba_mt", "num_bytes" : 93678, "name" : "validation" } }, "config_name" : "ces-deu", "size_in_bytes" : 376215, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 4615, "supervised_keys" : null }, "eng-mal" : { "supervised_keys" : null, "dataset_size" : 801, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-mal", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 801, "num_bytes" : 113410 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 113410 }, "dan-nob" : { "dataset_size" : 2309, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 180742, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 1011, "dataset_name" : "tatoeba_mt", "num_bytes" : 79099, "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1298, "num_bytes" : 101643, "name" : "test" } }, "config_name" : "dan-nob" }, "pol-cmn_Hans" : { "config_name" : "pol-cmn_Hans", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 367, "num_bytes" : 30094, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 30094, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 367 }, "fra-hbs" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 473, "builder_name" : "tatoeba_mt", "size_in_bytes" : 37837, "config_name" : "fra-hbs", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 473, "num_bytes" : 37837 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "fra-jpn_Hani" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 441070, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 3263, "num_bytes" : 324946 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1178, "num_bytes" : 116124 } }, "config_name" : "fra-jpn_Hani", "dataset_size" : 4441, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 } }, "deu-jbo" : { "supervised_keys" : null, "dataset_size" : 1448, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-jbo", "splits" : { "test" : { "name" : "test", "num_examples" : 1448, "dataset_name" : "tatoeba_mt", "num_bytes" : 114176 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 114176 }, "eng-vie" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 532628, "config_name" : "eng-vie", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 3134, "num_bytes" : 298384, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 234244, "num_examples" : 2499, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 5633, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "fra-pol" : { "supervised_keys" : null, "dataset_size" : 4090, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 341882, "config_name" : "fra-pol", "splits" : { "test" : { "name" : "test", "num_bytes" : 256127, "num_examples" : 3086, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 85755, "num_examples" : 1004, "dataset_name" : "tatoeba_mt", "name" : "validation" } } }, "jpn_Hira-pol" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 21773, "supervised_keys" : null, "size_in_bytes" : 2303630, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 1378699, "dataset_name" : "tatoeba_mt", "num_examples" : 13009 }, "test" : { "name" : "test", "num_bytes" : 924931, "num_examples" : 8764, "dataset_name" : "tatoeba_mt" } }, "config_name" : "jpn_Hira-pol", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "deu-isl" : { "config_name" : "deu-isl", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 968, "num_bytes" : 68421, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 68421, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 968 }, "kor-spa" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 939, "builder_name" : "tatoeba_mt", "size_in_bytes" : 83074, "config_name" : "kor-spa", "splits" : { "test" : { "name" : "test", "num_bytes" : 83074, "num_examples" : 939, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "por-rus" : { "dataset_size" : 20048, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_examples" : 9999, "dataset_name" : "tatoeba_mt", "num_bytes" : 1012258, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10049, "num_bytes" : 1048814, "name" : "validation" } }, "config_name" : "por-rus", "size_in_bytes" : 2061072, "builder_name" : "tatoeba_mt" }, "kab-kab" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 117278, "config_name" : "kab-kab", "splits" : { "validation" : { "num_bytes" : 24446, "dataset_name" : "tatoeba_mt", "num_examples" : 308, "name" : "validation" }, "test" : { "num_bytes" : 92832, "dataset_name" : "tatoeba_mt", "num_examples" : 992, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 1300, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "rus-tat" : { "dataset_size" : 3141, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 652494, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 2137, "dataset_name" : "tatoeba_mt", "num_bytes" : 448786 }, "validation" : { "num_examples" : 1004, "dataset_name" : "tatoeba_mt", "num_bytes" : 203708, "name" : "validation" } }, "config_name" : "rus-tat" }, "lat-ukr" : { "dataset_size" : 381, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 29975, "num_examples" : 381, "dataset_name" : "tatoeba_mt" } }, "config_name" : "lat-ukr", "size_in_bytes" : 29975, "builder_name" : "tatoeba_mt" }, "ara-jpn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 136439, "config_name" : "ara-jpn", "splits" : { "test" : { "name" : "test", "num_examples" : 1335, "dataset_name" : "tatoeba_mt", "num_bytes" : 136439 } }, "supervised_keys" : null, "dataset_size" : 1335, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-jpn_Kana" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 120011, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 829, "num_bytes" : 86656 }, "test" : { "name" : "test", "num_bytes" : 33355, "num_examples" : 320, "dataset_name" : "tatoeba_mt" } }, "config_name" : "deu-jpn_Kana", "dataset_size" : 1149, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "kab-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 39530, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 39530, "dataset_name" : "tatoeba_mt", "num_examples" : 422 } }, "config_name" : "kab-rus", "dataset_size" : 422, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "lfn_Latn-yid" : { "config_name" : "lfn_Latn-yid", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 18098, "num_examples" : 189, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 897, "num_bytes" : 81847 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 99945, "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1086 }, "deu-yid" : { "supervised_keys" : null, "dataset_size" : 972, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-yid", "splits" : { "validation" : { "num_bytes" : 13155, "num_examples" : 120, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 79442, "num_examples" : 852, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 92597 }, "tur-cmn_Hans" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 40855, "config_name" : "tur-cmn_Hans", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 584, "num_bytes" : 40855 } }, "supervised_keys" : null, "dataset_size" : 584, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0" }, "epo-hun" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 38848, "config_name" : "epo-hun", "splits" : { "test" : { "name" : "test", "num_bytes" : 843261, "dataset_name" : "tatoeba_mt", "num_examples" : 10093 }, "validation" : { "num_examples" : 28755, "dataset_name" : "tatoeba_mt", "num_bytes" : 2405986, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 3249247, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ces-pol" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 40371, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 40371, "dataset_name" : "tatoeba_mt", "num_examples" : 567 } }, "config_name" : "ces-pol", "dataset_size" : 567, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "ita-por" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 9435, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 6370, "num_bytes" : 513921 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 3065, "num_bytes" : 274134, "name" : "test" } }, "config_name" : "ita-por", "size_in_bytes" : 788055, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "cor-deu" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 47723, "config_name" : "cor-deu", "splits" : { "test" : { "name" : "test", "num_examples" : 820, "dataset_name" : "tatoeba_mt", "num_bytes" : 47723 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 820 }, "ita-nld" : { "supervised_keys" : null, "dataset_size" : 9516, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 642606, "config_name" : "ita-nld", "splits" : { "test" : { "num_examples" : 2577, "dataset_name" : "tatoeba_mt", "num_bytes" : 173198, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 469408, "dataset_name" : "tatoeba_mt", "num_examples" : 6939 } } }, "fra-tat" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 304, "builder_name" : "tatoeba_mt", "size_in_bytes" : 25778, "config_name" : "fra-tat", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 304, "num_bytes" : 25778, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ltz-nld" : { "config_name" : "ltz-nld", "splits" : { "test" : { "num_bytes" : 17279, "num_examples" : 291, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 17279, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 291 }, "hun-jpn_Hani" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 99344, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 62022, "num_examples" : 578, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 37322, "num_examples" : 350, "dataset_name" : "tatoeba_mt" } }, "config_name" : "hun-jpn_Hani", "dataset_size" : 928, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "dtp-jpn_Hira" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 235, "supervised_keys" : null, "splits" : { "test" : { "num_examples" : 235, "dataset_name" : "tatoeba_mt", "num_bytes" : 20048, "name" : "test" } }, "config_name" : "dtp-jpn_Hira", "size_in_bytes" : 20048, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn-nld" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 407659, "config_name" : "jpn-nld", "splits" : { "validation" : { "num_bytes" : 95717, "num_examples" : 1056, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 311942, "dataset_name" : "tatoeba_mt", "num_examples" : 3429 } }, "supervised_keys" : null, "dataset_size" : 4485, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "slv-zho" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 60602, "num_examples" : 824, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "slv-zho", "size_in_bytes" : 60602, "builder_name" : "tatoeba_mt", "dataset_size" : 824, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "jpn-por" : { "dataset_size" : 3071, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 193304, "dataset_name" : "tatoeba_mt", "num_examples" : 1939, "name" : "test" }, "validation" : { "num_bytes" : 113755, "num_examples" : 1132, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "jpn-por", "size_in_bytes" : 307059, "builder_name" : "tatoeba_mt" }, "fra-nor" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 33674, "config_name" : "fra-nor", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 476, "num_bytes" : 33674 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 476 }, "ell-spa" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 2832, "config_name" : "ell-spa", "splits" : { "test" : { "num_bytes" : 155748, "dataset_name" : "tatoeba_mt", "num_examples" : 1828, "name" : "test" }, "validation" : { "num_examples" : 1004, "dataset_name" : "tatoeba_mt", "num_bytes" : 85426, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 241174, "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "heb-por" : { "supervised_keys" : null, "dataset_size" : 741, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "heb-por", "splits" : { "test" : { "name" : "test", "num_bytes" : 64157, "dataset_name" : "tatoeba_mt", "num_examples" : 718 }, "validation" : { "name" : "validation", "num_bytes" : 1554, "dataset_name" : "tatoeba_mt", "num_examples" : 23 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 65711 }, "ces-fra" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 437, "config_name" : "ces-fra", "splits" : { "test" : { "num_bytes" : 34150, "dataset_name" : "tatoeba_mt", "num_examples" : 437, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 34150, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "hun-ita" : { "size_in_bytes" : 798061, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 7158, "dataset_name" : "tatoeba_mt", "num_bytes" : 468816, "name" : "validation" }, "test" : { "num_bytes" : 329245, "num_examples" : 4999, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "hun-ita", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 12157, "supervised_keys" : null }, "epo-tur" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 911833, "config_name" : "epo-tur", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 7649, "num_bytes" : 552160 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 4999, "num_bytes" : 359673, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 12648 }, "heb-nld" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 6855, "supervised_keys" : null, "splits" : { "test" : { "num_examples" : 2499, "dataset_name" : "tatoeba_mt", "num_bytes" : 236821, "name" : "test" }, "validation" : { "num_bytes" : 416293, "dataset_name" : "tatoeba_mt", "num_examples" : 4356, "name" : "validation" } }, "config_name" : "heb-nld", "size_in_bytes" : 653114, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "awa-eng" : { "supervised_keys" : null, "dataset_size" : 278, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "awa-eng", "splits" : { "test" : { "name" : "test", "num_bytes" : 20696, "dataset_name" : "tatoeba_mt", "num_examples" : 278 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 20696 }, "jbo_Latn-jpn_Hira" : { "supervised_keys" : null, "dataset_size" : 825, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "jbo_Latn-jpn_Hira", "splits" : { "test" : { "name" : "test", "num_examples" : 825, "dataset_name" : "tatoeba_mt", "num_bytes" : 72892 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 72892 }, "lad-yid" : { "supervised_keys" : null, "dataset_size" : 686, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 63743, "config_name" : "lad-yid", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 8610, "dataset_name" : "tatoeba_mt", "num_examples" : 83 }, "test" : { "num_bytes" : 55133, "num_examples" : 603, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "jpn_Hira-vie" : { "dataset_size" : 1429, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 166474, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 107208, "num_examples" : 927, "dataset_name" : "tatoeba_mt" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 502, "num_bytes" : 59266 } }, "config_name" : "jpn_Hira-vie" }, "epo-toki" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 346537, "config_name" : "epo-toki", "splits" : { "test" : { "num_bytes" : 234998, "num_examples" : 2732, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1294, "num_bytes" : 111539 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 4026 }, "por-zho" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 75621, "num_examples" : 1032, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "por-zho", "size_in_bytes" : 75621, "builder_name" : "tatoeba_mt", "dataset_size" : 1032, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "gla-spa" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 19141, "config_name" : "gla-spa", "splits" : { "test" : { "num_examples" : 288, "dataset_name" : "tatoeba_mt", "num_bytes" : 19141, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 288 }, "deu-ukr" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 1835865, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 971543, "num_examples" : 11720, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10318, "num_bytes" : 864322, "name" : "test" } }, "config_name" : "deu-ukr", "dataset_size" : 22038, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 } }, "pol-tur" : { "dataset_size" : 891, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 66267, "dataset_name" : "tatoeba_mt", "num_examples" : 891, "name" : "test" } }, "config_name" : "pol-tur", "size_in_bytes" : 66267, "builder_name" : "tatoeba_mt" }, "epo-nob" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 4077, "config_name" : "epo-nob", "splits" : { "test" : { "num_examples" : 3091, "dataset_name" : "tatoeba_mt", "num_bytes" : 308406, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 98483, "num_examples" : 986, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 406889, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "tlh_Latn-cmn_Hans" : { "supervised_keys" : null, "dataset_size" : 245, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "tlh_Latn-cmn_Hans", "splits" : { "test" : { "name" : "test", "num_bytes" : 17590, "num_examples" : 245, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 17590 }, "fra-fra" : { "supervised_keys" : null, "dataset_size" : 3116, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 258985, "config_name" : "fra-fra", "splits" : { "test" : { "num_examples" : 999, "dataset_name" : "tatoeba_mt", "num_bytes" : 84132, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 2117, "dataset_name" : "tatoeba_mt", "num_bytes" : 174853 } } }, "lat-yid" : { "dataset_size" : 534, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_examples" : 457, "dataset_name" : "tatoeba_mt", "num_bytes" : 41312, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 77, "num_bytes" : 6791 } }, "config_name" : "lat-yid", "size_in_bytes" : 48103, "builder_name" : "tatoeba_mt" }, "bel-nld" : { "size_in_bytes" : 62683, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 605, "dataset_name" : "tatoeba_mt", "num_bytes" : 62683 } }, "config_name" : "bel-nld", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 605, "supervised_keys" : null }, "cor-ita" : { "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "cor-ita", "splits" : { "test" : { "name" : "test", "num_bytes" : 14528, "num_examples" : 286, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 14528, "supervised_keys" : null, "dataset_size" : 286, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-ltz" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 346, "config_name" : "deu-ltz", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 346, "num_bytes" : 23796 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 23796, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "deu-epo" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 23684146, "config_name" : "deu-epo", "splits" : { "validation" : { "name" : "validation", "num_examples" : 221680, "dataset_name" : "tatoeba_mt", "num_bytes" : 22254500 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 13438, "num_bytes" : 1429646, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 235118 }, "cbk-eng" : { "supervised_keys" : null, "dataset_size" : 2497, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "cbk-eng", "splits" : { "validation" : { "name" : "validation", "num_examples" : 1000, "dataset_name" : "tatoeba_mt", "num_bytes" : 73581 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1497, "num_bytes" : 110361 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 183942 }, "hin-urd" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 33691, "config_name" : "hin-urd", "splits" : { "test" : { "name" : "test", "num_bytes" : 33691, "dataset_name" : "tatoeba_mt", "num_examples" : 239 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 239 }, "fra-nds" : { "supervised_keys" : null, "dataset_size" : 856, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fra-nds", "splits" : { "test" : { "name" : "test", "num_examples" : 856, "dataset_name" : "tatoeba_mt", "num_bytes" : 59528 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 59528 }, "fin-ita" : { "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fin-ita", "splits" : { "validation" : { "num_bytes" : 69745, "num_examples" : 1002, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1038, "num_bytes" : 72837, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 142582, "supervised_keys" : null, "dataset_size" : 2040, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "fra-zho" : { "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fra-zho", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 9993, "num_bytes" : 896410 }, "validation" : { "num_bytes" : 674827, "dataset_name" : "tatoeba_mt", "num_examples" : 7557, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1571237, "supervised_keys" : null, "dataset_size" : 17550, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "kor_Hang-cmn_Hans" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 18974, "dataset_name" : "tatoeba_mt", "num_examples" : 206 } }, "config_name" : "kor_Hang-cmn_Hans", "size_in_bytes" : 18974, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 206, "supervised_keys" : null }, "eng-hun" : { "dataset_size" : 110179, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 959274, "dataset_name" : "tatoeba_mt", "num_examples" : 13036, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 97143, "num_bytes" : 6949776 } }, "config_name" : "eng-hun", "size_in_bytes" : 7909050, "builder_name" : "tatoeba_mt" }, "lad_Latn-spa" : { "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "lad_Latn-spa", "splits" : { "test" : { "name" : "test", "num_bytes" : 14299, "num_examples" : 238, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 14299, "supervised_keys" : null, "dataset_size" : 238, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "ina-tur" : { "supervised_keys" : null, "dataset_size" : 316, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 20364, "config_name" : "ina-tur", "splits" : { "test" : { "num_examples" : 316, "dataset_name" : "tatoeba_mt", "num_bytes" : 20364, "name" : "test" } } }, "eng-uzb_Latn" : { "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-uzb_Latn", "splits" : { "test" : { "name" : "test", "num_examples" : 300, "dataset_name" : "tatoeba_mt", "num_bytes" : 17949 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 17949, "supervised_keys" : null, "dataset_size" : 300, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "kaz-rus" : { "config_name" : "kaz-rus", "splits" : { "validation" : { "name" : "validation", "num_examples" : 1015, "dataset_name" : "tatoeba_mt", "num_bytes" : 241995 }, "test" : { "name" : "test", "num_bytes" : 337480, "dataset_name" : "tatoeba_mt", "num_examples" : 1399 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 579475, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 2414 }, "afr-deu" : { "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "afr-deu", "splits" : { "test" : { "name" : "test", "num_examples" : 1582, "dataset_name" : "tatoeba_mt", "num_bytes" : 99941 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 99941, "supervised_keys" : null, "dataset_size" : 1582, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "epo-vie" : { "splits" : { "test" : { "num_bytes" : 173445, "num_examples" : 1789, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "epo-vie", "size_in_bytes" : 173445, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 1789, "supervised_keys" : null }, "chv-eng" : { "config_name" : "chv-eng", "splits" : { "test" : { "name" : "test", "num_examples" : 335, "dataset_name" : "tatoeba_mt", "num_bytes" : 28676 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 28676, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 335 }, "ber-spa" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 24058, "supervised_keys" : null, "size_in_bytes" : 1938442, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10015, "num_bytes" : 807626, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 14043, "num_bytes" : 1130816 } }, "config_name" : "ber-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "dan-eng" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 2299179, "config_name" : "dan-eng", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 20088, "num_bytes" : 1491198 }, "test" : { "name" : "test", "num_bytes" : 807981, "num_examples" : 10794, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 30882 }, "rus-zho" : { "supervised_keys" : null, "dataset_size" : 7342, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "rus-zho", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 260121, "name" : "test" }, "validation" : { "num_bytes" : 504643, "dataset_name" : "tatoeba_mt", "num_examples" : 4843, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 764764 }, "jpn_Hira-tlh_Latn" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 62553, "config_name" : "jpn_Hira-tlh_Latn", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 637, "num_bytes" : 62553 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 637 }, "eng-nob" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 9739, "config_name" : "eng-nob", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 5201, "num_bytes" : 420251, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 370388, "dataset_name" : "tatoeba_mt", "num_examples" : 4538 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 790639, "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn_Hira-nld" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "num_bytes" : 82569, "num_examples" : 894, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 269634, "dataset_name" : "tatoeba_mt", "num_examples" : 2924, "name" : "test" } }, "config_name" : "jpn_Hira-nld", "size_in_bytes" : 352203, "builder_name" : "tatoeba_mt", "dataset_size" : 3818, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "cha-eng" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 11634, "config_name" : "cha-eng", "splits" : { "test" : { "name" : "test", "num_examples" : 226, "dataset_name" : "tatoeba_mt", "num_bytes" : 11634 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 226 }, "fra-rus" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 183049, "num_bytes" : 17211453, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 11489, "num_bytes" : 1073575 } }, "config_name" : "fra-rus", "size_in_bytes" : 18285028, "builder_name" : "tatoeba_mt", "dataset_size" : 194538, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "eng-swa" : { "dataset_size" : 386, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 22845, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 386, "dataset_name" : "tatoeba_mt", "num_bytes" : 22845 } }, "config_name" : "eng-swa" }, "fra-kab" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 30852, "config_name" : "fra-kab", "splits" : { "test" : { "name" : "test", "num_examples" : 12490, "dataset_name" : "tatoeba_mt", "num_bytes" : 1051694 }, "validation" : { "num_examples" : 18362, "dataset_name" : "tatoeba_mt", "num_bytes" : 1511811, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2563505, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ita-ind" : { "config_name" : "ita-ind", "splits" : { "test" : { "name" : "test", "num_bytes" : 27058, "dataset_name" : "tatoeba_mt", "num_examples" : 367 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 27058, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 367 }, "dtp-msa" : { "dataset_size" : 516, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 53688, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 53688, "dataset_name" : "tatoeba_mt", "num_examples" : 516, "name" : "test" } }, "config_name" : "dtp-msa" }, "eng-mlt" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 202, "config_name" : "eng-mlt", "splits" : { "test" : { "name" : "test", "num_examples" : 202, "dataset_name" : "tatoeba_mt", "num_bytes" : 11648 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 11648, "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-tur" : { "splits" : { "test" : { "name" : "test", "num_examples" : 13906, "dataset_name" : "tatoeba_mt", "num_bytes" : 1133690 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 658319, "num_bytes" : 52917848, "name" : "validation" } }, "config_name" : "eng-tur", "size_in_bytes" : 54051538, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 672225, "supervised_keys" : null }, "ain-fin" : { "supervised_keys" : null, "dataset_size" : 205, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 14201, "config_name" : "ain-fin", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 205, "num_bytes" : 14201, "name" : "test" } } }, "tlh-zho" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 31164, "dataset_name" : "tatoeba_mt", "num_examples" : 447 } }, "config_name" : "tlh-zho", "size_in_bytes" : 31164, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 447, "supervised_keys" : null }, "gos-nld" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 2277, "supervised_keys" : null, "size_in_bytes" : 126574, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1851, "num_bytes" : 100409 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 426, "num_bytes" : 26165 } }, "config_name" : "gos-nld", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "fra-lfn_Latn" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 39268, "config_name" : "fra-lfn_Latn", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 59, "num_bytes" : 6683 }, "test" : { "name" : "test", "num_bytes" : 32585, "num_examples" : 360, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 419 }, "ces-rus" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 8398, "config_name" : "ces-rus", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 474691, "dataset_name" : "tatoeba_mt", "num_examples" : 5465 }, "test" : { "num_bytes" : 254127, "dataset_name" : "tatoeba_mt", "num_examples" : 2933, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 728818, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "est-rus" : { "supervised_keys" : null, "dataset_size" : 612, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 52902, "config_name" : "est-rus", "splits" : { "test" : { "num_examples" : 612, "dataset_name" : "tatoeba_mt", "num_bytes" : 52902, "name" : "test" } } }, "nld-ukr" : { "dataset_size" : 15063, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 775813, "num_examples" : 9999, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "num_bytes" : 392653, "num_examples" : 5064, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "nld-ukr", "size_in_bytes" : 1168466, "builder_name" : "tatoeba_mt" }, "eng-toki" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 401418, "num_examples" : 4989, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 709290, "dataset_name" : "tatoeba_mt", "num_examples" : 8701 } }, "config_name" : "eng-toki", "size_in_bytes" : 1110708, "builder_name" : "tatoeba_mt", "dataset_size" : 13690, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "rus-rus" : { "supervised_keys" : null, "dataset_size" : 3904, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 502261, "config_name" : "rus-rus", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 312967 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1405, "num_bytes" : 189294 } } }, "fin-jpn_Hani" : { "supervised_keys" : null, "dataset_size" : 2082, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 165279, "config_name" : "fin-jpn_Hani", "splits" : { "test" : { "name" : "test", "num_bytes" : 74484, "num_examples" : 975, "dataset_name" : "tatoeba_mt" }, "validation" : { "name" : "validation", "num_bytes" : 90795, "num_examples" : 1107, "dataset_name" : "tatoeba_mt" } } }, "swe-cmn_Hant" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 566, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_examples" : 566, "dataset_name" : "tatoeba_mt", "num_bytes" : 35044 } }, "config_name" : "swe-cmn_Hant", "size_in_bytes" : 35044, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "deu-msa" : { "dataset_size" : 543, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_examples" : 543, "dataset_name" : "tatoeba_mt", "num_bytes" : 46985, "name" : "test" } }, "config_name" : "deu-msa", "size_in_bytes" : 46985, "builder_name" : "tatoeba_mt" }, "jpn_Hira-swe" : { "dataset_size" : 728, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 728, "num_bytes" : 70646 } }, "config_name" : "jpn_Hira-swe", "size_in_bytes" : 70646, "builder_name" : "tatoeba_mt" }, "fin-pol" : { "supervised_keys" : null, "dataset_size" : 608, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 43252, "config_name" : "fin-pol", "splits" : { "test" : { "num_bytes" : 43252, "num_examples" : 608, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "epo-lit" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 1652313, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10428, "num_bytes" : 772441, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 11510, "dataset_name" : "tatoeba_mt", "num_bytes" : 879872 } }, "config_name" : "epo-lit", "dataset_size" : 21938, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "eus-jpn" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 236, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_examples" : 236, "dataset_name" : "tatoeba_mt", "num_bytes" : 14683 } }, "config_name" : "eus-jpn", "size_in_bytes" : 14683, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "hun-rus" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 8783, "builder_name" : "tatoeba_mt", "size_in_bytes" : 818123, "config_name" : "hun-rus", "splits" : { "test" : { "name" : "test", "num_bytes" : 252859, "num_examples" : 2686, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_examples" : 6097, "dataset_name" : "tatoeba_mt", "num_bytes" : 565264, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "eng-lfn" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 7001, "supervised_keys" : null, "size_in_bytes" : 597752, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 334527, "num_examples" : 3705, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_examples" : 3296, "dataset_name" : "tatoeba_mt", "num_bytes" : 263225 } }, "config_name" : "eng-lfn", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } } }, "fry-nld" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fry-nld", "splits" : { "test" : { "num_examples" : 259, "dataset_name" : "tatoeba_mt", "num_bytes" : 18557, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 18557, "supervised_keys" : null, "dataset_size" : 259, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-mya" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-mya", "splits" : { "test" : { "num_examples" : 215, "dataset_name" : "tatoeba_mt", "num_bytes" : 34381, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 34381, "supervised_keys" : null, "dataset_size" : 215, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "fra-run" : { "supervised_keys" : null, "dataset_size" : 1273, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fra-run", "splits" : { "test" : { "name" : "test", "num_bytes" : 72697, "num_examples" : 1273, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 72697 }, "ell-epo" : { "config_name" : "ell-epo", "splits" : { "test" : { "name" : "test", "num_bytes" : 58730, "dataset_name" : "tatoeba_mt", "num_examples" : 603 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 58730, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 603 }, "jbo_Latn-cmn_Hans" : { "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "jbo_Latn-cmn_Hans", "splits" : { "test" : { "name" : "test", "num_examples" : 280, "dataset_name" : "tatoeba_mt", "num_bytes" : 20111 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 20111, "supervised_keys" : null, "dataset_size" : 280, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-heb" : { "dataset_size" : 164088, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 14079385, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 153570, "num_bytes" : 13176238 }, "test" : { "num_bytes" : 903147, "dataset_name" : "tatoeba_mt", "num_examples" : 10518, "name" : "test" } }, "config_name" : "eng-heb" }, "epo-ina" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 376998, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 2643, "dataset_name" : "tatoeba_mt", "num_bytes" : 250841, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 126157, "num_examples" : 1447, "dataset_name" : "tatoeba_mt" } }, "config_name" : "epo-ina", "dataset_size" : 4090, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "jpn_Hani-pol" : { "config_name" : "jpn_Hani-pol", "splits" : { "validation" : { "name" : "validation", "num_examples" : 1570, "dataset_name" : "tatoeba_mt", "num_bytes" : 151313 }, "test" : { "name" : "test", "num_bytes" : 104212, "num_examples" : 1080, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 255525, "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2650 }, "deu-spa" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 85580, "config_name" : "deu-spa", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 75060, "num_bytes" : 6578781, "name" : "validation" }, "test" : { "num_examples" : 10520, "dataset_name" : "tatoeba_mt", "num_bytes" : 930791, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 7509572, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "cor-rus" : { "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "cor-rus", "splits" : { "test" : { "name" : "test", "num_bytes" : 14472, "dataset_name" : "tatoeba_mt", "num_examples" : 217 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 14472, "supervised_keys" : null, "dataset_size" : 217, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-lad" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 52191, "config_name" : "eng-lad", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 767, "num_bytes" : 48380 }, "validation" : { "name" : "validation", "num_bytes" : 3811, "num_examples" : 57, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 824, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-yue_Hant" : { "supervised_keys" : null, "dataset_size" : 1974, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-yue_Hant", "splits" : { "validation" : { "num_bytes" : 113290, "dataset_name" : "tatoeba_mt", "num_examples" : 1521, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 453, "num_bytes" : 35124 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 148414 }, "heb-cmn_Hant" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 328, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 328, "num_bytes" : 23289 } }, "config_name" : "heb-cmn_Hant", "size_in_bytes" : 23289, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-hin" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 1251920, "config_name" : "eng-hin", "splits" : { "test" : { "name" : "test", "num_bytes" : 572488, "num_examples" : 4999, "dataset_name" : "tatoeba_mt" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 5943, "num_bytes" : 679432, "name" : "validation" } }, "supervised_keys" : null, "dataset_size" : 10942, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "hun-zho" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 33218, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 33218, "num_examples" : 433, "dataset_name" : "tatoeba_mt" } }, "config_name" : "hun-zho", "dataset_size" : 433, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "avk-fra" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1285, "config_name" : "avk-fra", "splits" : { "test" : { "num_bytes" : 107793, "dataset_name" : "tatoeba_mt", "num_examples" : 1243, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 3230, "dataset_name" : "tatoeba_mt", "num_examples" : 42 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 111023, "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fin-nor" : { "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fin-nor", "splits" : { "test" : { "num_bytes" : 160613, "dataset_name" : "tatoeba_mt", "num_examples" : 2487, "name" : "test" }, "validation" : { "num_examples" : 1147, "dataset_name" : "tatoeba_mt", "num_bytes" : 75759, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 236372, "supervised_keys" : null, "dataset_size" : 3634, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "ina-lad" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 26848, "config_name" : "ina-lad", "splits" : { "validation" : { "num_bytes" : 3378, "dataset_name" : "tatoeba_mt", "num_examples" : 45, "name" : "validation" }, "test" : { "num_examples" : 319, "dataset_name" : "tatoeba_mt", "num_bytes" : 23470, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 364, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "msa-spa" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 16593, "num_examples" : 231, "dataset_name" : "tatoeba_mt" } }, "config_name" : "msa-spa", "size_in_bytes" : 16593, "builder_name" : "tatoeba_mt", "dataset_size" : 231, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 } }, "fra-ron" : { "supervised_keys" : null, "dataset_size" : 1924, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 140669, "config_name" : "fra-ron", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1924, "num_bytes" : 140669 } } }, "deu-tgl" : { "dataset_size" : 325, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 28456, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 325, "dataset_name" : "tatoeba_mt", "num_bytes" : 28456 } }, "config_name" : "deu-tgl" }, "lat-spa" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 461356, "config_name" : "lat-spa", "splits" : { "test" : { "num_bytes" : 346603, "dataset_name" : "tatoeba_mt", "num_examples" : 3128, "name" : "test" }, "validation" : { "num_bytes" : 114753, "num_examples" : 1067, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 4195 }, "ita-cmn_Hant" : { "splits" : { "test" : { "num_bytes" : 72270, "dataset_name" : "tatoeba_mt", "num_examples" : 1036, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 397, "dataset_name" : "tatoeba_mt", "num_bytes" : 28675 } }, "config_name" : "ita-cmn_Hant", "size_in_bytes" : 100945, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 1433, "supervised_keys" : null }, "ceb-deu" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 76229, "config_name" : "ceb-deu", "splits" : { "test" : { "num_bytes" : 76229, "num_examples" : 902, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 902 }, "fin-fra" : { "dataset_size" : 2974, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 203632, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1919, "num_bytes" : 131752 }, "validation" : { "name" : "validation", "num_bytes" : 71880, "num_examples" : 1055, "dataset_name" : "tatoeba_mt" } }, "config_name" : "fin-fra" }, "lfn-por" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 2940, "supervised_keys" : null, "size_in_bytes" : 415994, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 1944, "dataset_name" : "tatoeba_mt", "num_bytes" : 280715, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 996, "dataset_name" : "tatoeba_mt", "num_bytes" : 135279 } }, "config_name" : "lfn-por", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "lad-spa" : { "dataset_size" : 299, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 19271, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 1938, "dataset_name" : "tatoeba_mt", "num_examples" : 24 }, "test" : { "name" : "test", "num_examples" : 275, "dataset_name" : "tatoeba_mt", "num_bytes" : 17333 } }, "config_name" : "lad-spa" }, "ina-lfn" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 889, "builder_name" : "tatoeba_mt", "size_in_bytes" : 64972, "config_name" : "ina-lfn", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 9054, "num_examples" : 121, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 55918, "num_examples" : 768, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "spa-swe" : { "size_in_bytes" : 104137, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 1350, "dataset_name" : "tatoeba_mt", "num_bytes" : 89887, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 14250, "num_examples" : 203, "dataset_name" : "tatoeba_mt" } }, "config_name" : "spa-swe", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 1553, "supervised_keys" : null }, "deu-eus" : { "supervised_keys" : null, "dataset_size" : 455, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-eus", "splits" : { "test" : { "num_bytes" : 35688, "dataset_name" : "tatoeba_mt", "num_examples" : 455, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 35688 }, "srp_Latn-ukr" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "srp_Latn-ukr", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 347, "num_bytes" : 25013 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 25013, "supervised_keys" : null, "dataset_size" : 347, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "cat-nld" : { "size_in_bytes" : 41913, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 577, "num_bytes" : 41913 } }, "config_name" : "cat-nld", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 577, "supervised_keys" : null }, "msa-msa" : { "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "msa-msa", "splits" : { "test" : { "name" : "test", "num_bytes" : 81386, "dataset_name" : "tatoeba_mt", "num_examples" : 957 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 81386, "supervised_keys" : null, "dataset_size" : 957, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-cmn_Hans" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 1947321, "config_name" : "eng-cmn_Hans", "splits" : { "validation" : { "num_bytes" : 1561687, "dataset_name" : "tatoeba_mt", "num_examples" : 17966, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 385634, "dataset_name" : "tatoeba_mt", "num_examples" : 4449 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 22415 }, "ron-tur" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 3468, "builder_name" : "tatoeba_mt", "size_in_bytes" : 286684, "config_name" : "ron-tur", "splits" : { "test" : { "name" : "test", "num_examples" : 2459, "dataset_name" : "tatoeba_mt", "num_bytes" : 203717 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1009, "num_bytes" : 82967, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "cat-por" : { "supervised_keys" : null, "dataset_size" : 746, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "cat-por", "splits" : { "test" : { "name" : "test", "num_bytes" : 65694, "dataset_name" : "tatoeba_mt", "num_examples" : 746 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 65694 }, "nno-nob" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_examples" : 466, "dataset_name" : "tatoeba_mt", "num_bytes" : 31927, "name" : "test" } }, "config_name" : "nno-nob", "size_in_bytes" : 31927, "builder_name" : "tatoeba_mt", "dataset_size" : 466, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "eng-fao" : { "dataset_size" : 293, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 20156, "dataset_name" : "tatoeba_mt", "num_examples" : 293, "name" : "test" } }, "config_name" : "eng-fao", "size_in_bytes" : 20156, "builder_name" : "tatoeba_mt" }, "epo-ukr" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 7057, "builder_name" : "tatoeba_mt", "size_in_bytes" : 571357, "config_name" : "epo-ukr", "splits" : { "test" : { "name" : "test", "num_bytes" : 201989, "dataset_name" : "tatoeba_mt", "num_examples" : 2499 }, "validation" : { "name" : "validation", "num_examples" : 4558, "dataset_name" : "tatoeba_mt", "num_bytes" : 369368 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "deu-nob" : { "dataset_size" : 4489, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "num_bytes" : 97801, "dataset_name" : "tatoeba_mt", "num_examples" : 965, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 363529, "dataset_name" : "tatoeba_mt", "num_examples" : 3524 } }, "config_name" : "deu-nob", "size_in_bytes" : 461330, "builder_name" : "tatoeba_mt" }, "eng-khm" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_examples" : 725, "dataset_name" : "tatoeba_mt", "num_bytes" : 71191, "name" : "test" } }, "config_name" : "eng-khm", "size_in_bytes" : 71191, "builder_name" : "tatoeba_mt", "dataset_size" : 725, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 } }, "afr-nld" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 1055, "supervised_keys" : null, "splits" : { "test" : { "num_examples" : 1055, "dataset_name" : "tatoeba_mt", "num_bytes" : 68060, "name" : "test" } }, "config_name" : "afr-nld", "size_in_bytes" : 68060, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "vie-vie" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 68696, "config_name" : "vie-vie", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 541, "num_bytes" : 68696 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 541 }, "jpn_Kana-spa" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 691, "supervised_keys" : null, "size_in_bytes" : 69865, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 24577, "dataset_name" : "tatoeba_mt", "num_examples" : 253, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 438, "num_bytes" : 45288, "name" : "validation" } }, "config_name" : "jpn_Kana-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "dan-jpn_Hira" : { "dataset_size" : 715, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 59987, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 59987, "num_examples" : 715, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "dan-jpn_Hira" }, "por-swe" : { "dataset_size" : 319, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 21549, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 21549, "num_examples" : 319, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "por-swe" }, "ita-ron" : { "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1004, "num_bytes" : 68917 } }, "config_name" : "ita-ron", "size_in_bytes" : 68917, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 1004, "supervised_keys" : null }, "epo-epo" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 19674, "supervised_keys" : null, "size_in_bytes" : 1578599, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 9999, "num_bytes" : 797530 }, "validation" : { "num_examples" : 9675, "dataset_name" : "tatoeba_mt", "num_bytes" : 781069, "name" : "validation" } }, "config_name" : "epo-epo", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "fra-cmn_Hant" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 139241, "dataset_name" : "tatoeba_mt", "num_examples" : 1716 }, "test" : { "num_bytes" : 174852, "num_examples" : 2131, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "fra-cmn_Hant", "size_in_bytes" : 314093, "builder_name" : "tatoeba_mt", "dataset_size" : 3847, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "rus-cmn_Hant" : { "dataset_size" : 2322, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 218371, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 798, "num_bytes" : 74745, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 143626, "dataset_name" : "tatoeba_mt", "num_examples" : 1524 } }, "config_name" : "rus-cmn_Hant" }, "ell-eng" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 2061127, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 12919, "num_bytes" : 1097718 }, "test" : { "name" : "test", "num_examples" : 10898, "dataset_name" : "tatoeba_mt", "num_bytes" : 963409 } }, "config_name" : "ell-eng", "dataset_size" : 23817, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "pol-ukr" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 9415, "supervised_keys" : null, "splits" : { "test" : { "num_examples" : 2518, "dataset_name" : "tatoeba_mt", "num_bytes" : 199615, "name" : "test" }, "validation" : { "num_examples" : 6897, "dataset_name" : "tatoeba_mt", "num_bytes" : 550111, "name" : "validation" } }, "config_name" : "pol-ukr", "size_in_bytes" : 749726, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ara-ber" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "ara-ber", "splits" : { "test" : { "num_examples" : 2316, "dataset_name" : "tatoeba_mt", "num_bytes" : 209878, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 1031, "dataset_name" : "tatoeba_mt", "num_bytes" : 95068 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 304946, "supervised_keys" : null, "dataset_size" : 3347, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-tur" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1003877, "config_name" : "deu-tur", "splits" : { "validation" : { "num_examples" : 7276, "dataset_name" : "tatoeba_mt", "num_bytes" : 593893, "name" : "validation" }, "test" : { "num_bytes" : 409984, "num_examples" : 4999, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 12275, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "tur-uig" : { "supervised_keys" : null, "dataset_size" : 2409, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 259792, "config_name" : "tur-uig", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 109693, "dataset_name" : "tatoeba_mt", "num_examples" : 1010 }, "test" : { "name" : "test", "num_bytes" : 150099, "dataset_name" : "tatoeba_mt", "num_examples" : 1399 } } }, "eng-ilo" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 2091, "supervised_keys" : null, "size_in_bytes" : 150100, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 71696, "dataset_name" : "tatoeba_mt", "num_examples" : 999, "name" : "validation" }, "test" : { "num_examples" : 1092, "dataset_name" : "tatoeba_mt", "num_bytes" : 78404, "name" : "test" } }, "config_name" : "eng-ilo", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "deu-toki" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 23929, "config_name" : "deu-toki", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10113, "num_bytes" : 945160, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 1278234, "num_examples" : 13816, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2223394, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn_Hira-tur" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 71606, "config_name" : "jpn_Hira-tur", "splits" : { "test" : { "name" : "test", "num_examples" : 731, "dataset_name" : "tatoeba_mt", "num_bytes" : 71606 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 731 }, "eng-que" : { "supervised_keys" : null, "dataset_size" : 256, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 15806, "config_name" : "eng-que", "splits" : { "test" : { "name" : "test", "num_examples" : 256, "dataset_name" : "tatoeba_mt", "num_bytes" : 15806 } } }, "eng-lav" : { "dataset_size" : 1630, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 114179, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1630, "num_bytes" : 114179, "name" : "test" } }, "config_name" : "eng-lav" }, "deu-hun" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 69423, "builder_name" : "tatoeba_mt", "size_in_bytes" : 6049188, "config_name" : "deu-hun", "splits" : { "validation" : { "name" : "validation", "num_examples" : 54082, "dataset_name" : "tatoeba_mt", "num_bytes" : 4698144 }, "test" : { "name" : "test", "num_bytes" : 1351044, "num_examples" : 15341, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "eng-tzl" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 219, "supervised_keys" : null, "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 19, "num_bytes" : 1313, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 200, "num_bytes" : 10369 } }, "config_name" : "eng-tzl", "size_in_bytes" : 11682, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ita-pms" : { "config_name" : "ita-pms", "splits" : { "test" : { "name" : "test", "num_examples" : 231, "dataset_name" : "tatoeba_mt", "num_bytes" : 18864 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 18864, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 231 }, "hbs-ukr" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 941, "builder_name" : "tatoeba_mt", "size_in_bytes" : 76362, "config_name" : "hbs-ukr", "splits" : { "test" : { "num_bytes" : 76362, "num_examples" : 941, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "lit-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 811066, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 482377, "num_examples" : 5229, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 328689, "num_examples" : 3597, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "lit-rus", "dataset_size" : 8826, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "fin-fin" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1347, "builder_name" : "tatoeba_mt", "size_in_bytes" : 90801, "config_name" : "fin-fin", "splits" : { "test" : { "name" : "test", "num_bytes" : 66612, "dataset_name" : "tatoeba_mt", "num_examples" : 999 }, "validation" : { "name" : "validation", "num_examples" : 348, "dataset_name" : "tatoeba_mt", "num_bytes" : 24189 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "epo-jbo" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "epo-jbo", "splits" : { "test" : { "num_bytes" : 83091, "dataset_name" : "tatoeba_mt", "num_examples" : 1166, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 83091, "supervised_keys" : null, "dataset_size" : 1166, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "fin-por" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 31930, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 476, "dataset_name" : "tatoeba_mt", "num_bytes" : 31930 } }, "config_name" : "fin-por", "dataset_size" : 476, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "jpn_Hira-por" : { "dataset_size" : 2446, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 907, "num_bytes" : 93140, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 157262, "dataset_name" : "tatoeba_mt", "num_examples" : 1539 } }, "config_name" : "jpn_Hira-por", "size_in_bytes" : 250402, "builder_name" : "tatoeba_mt" }, "eng-nst" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 804, "supervised_keys" : null, "size_in_bytes" : 61426, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 61426, "num_examples" : 804, "dataset_name" : "tatoeba_mt" } }, "config_name" : "eng-nst", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } } }, "fin-nld" : { "size_in_bytes" : 21320, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 316, "dataset_name" : "tatoeba_mt", "num_bytes" : 21320, "name" : "test" } }, "config_name" : "fin-nld", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 316, "supervised_keys" : null }, "rus-yue_Hans" : { "supervised_keys" : null, "dataset_size" : 541, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 55380, "config_name" : "rus-yue_Hans", "splits" : { "validation" : { "num_bytes" : 31847, "num_examples" : 318, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 223, "num_bytes" : 23533, "name" : "test" } } }, "epo-isl" : { "dataset_size" : 232, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 17137, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 17137, "dataset_name" : "tatoeba_mt", "num_examples" : 232, "name" : "test" } }, "config_name" : "epo-isl" }, "eng-orv" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 26658, "config_name" : "eng-orv", "splits" : { "test" : { "name" : "test", "num_examples" : 321, "dataset_name" : "tatoeba_mt", "num_bytes" : 26658 } }, "supervised_keys" : null, "dataset_size" : 321, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "ara-rus" : { "dataset_size" : 4922, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 366839, "num_examples" : 3714, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 122685, "num_examples" : 1208, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "ara-rus", "size_in_bytes" : 489524, "builder_name" : "tatoeba_mt" }, "epo-yid" : { "config_name" : "epo-yid", "splits" : { "validation" : { "num_bytes" : 196929, "num_examples" : 2023, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 992, "dataset_name" : "tatoeba_mt", "num_bytes" : 94278 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 291207, "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 3015 }, "nor-spa" : { "supervised_keys" : null, "dataset_size" : 959, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 79106, "config_name" : "nor-spa", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 959, "num_bytes" : 79106 } } }, "tlh_Latn-yid" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 287, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_bytes" : 23118, "dataset_name" : "tatoeba_mt", "num_examples" : 287 } }, "config_name" : "tlh_Latn-yid", "size_in_bytes" : 23118, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ron-spa" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 1958, "supervised_keys" : null, "size_in_bytes" : 137180, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 137180, "dataset_name" : "tatoeba_mt", "num_examples" : 1958 } }, "config_name" : "ron-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "cat-fra" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 58832, "config_name" : "cat-fra", "splits" : { "test" : { "num_examples" : 699, "dataset_name" : "tatoeba_mt", "num_bytes" : 58832, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 699, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "ara-ell" : { "supervised_keys" : null, "dataset_size" : 425, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "ara-ell", "splits" : { "test" : { "num_bytes" : 32889, "dataset_name" : "tatoeba_mt", "num_examples" : 425, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 32889 }, "spa-tat" : { "dataset_size" : 440, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 30965, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 30965, "num_examples" : 440, "dataset_name" : "tatoeba_mt" } }, "config_name" : "spa-tat" }, "isl-jpn_Hira" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_examples" : 210, "dataset_name" : "tatoeba_mt", "num_bytes" : 18260 } }, "config_name" : "isl-jpn_Hira", "size_in_bytes" : 18260, "builder_name" : "tatoeba_mt", "dataset_size" : 210, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "bel-zho" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 324, "supervised_keys" : null, "size_in_bytes" : 29413, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 324, "dataset_name" : "tatoeba_mt", "num_bytes" : 29413 } }, "config_name" : "bel-zho", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } } }, "dan-epo" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 30303, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2319519, "config_name" : "dan-epo", "splits" : { "test" : { "name" : "test", "num_bytes" : 843351, "num_examples" : 11068, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 1476168, "num_examples" : 19235, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "eng-pam" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 493, "num_bytes" : 37820, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 75041, "dataset_name" : "tatoeba_mt", "num_examples" : 999 } }, "config_name" : "eng-pam", "size_in_bytes" : 112861, "builder_name" : "tatoeba_mt", "dataset_size" : 1492, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "epo-jpn_Hira" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 8559, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 289591, "num_examples" : 2985, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 538946, "dataset_name" : "tatoeba_mt", "num_examples" : 5574 } }, "config_name" : "epo-jpn_Hira", "size_in_bytes" : 828537, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ita-rus" : { "config_name" : "ita-rus", "splits" : { "test" : { "name" : "test", "num_bytes" : 994450, "dataset_name" : "tatoeba_mt", "num_examples" : 10044 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 66132, "num_bytes" : 6559508 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 7553958, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 76176 }, "epo-tlh" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 2092, "supervised_keys" : null, "size_in_bytes" : 137349, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 163, "num_bytes" : 9601 }, "test" : { "num_bytes" : 127748, "dataset_name" : "tatoeba_mt", "num_examples" : 1929, "name" : "test" } }, "config_name" : "epo-tlh", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "eng-tgl" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 589791, "config_name" : "eng-tgl", "splits" : { "validation" : { "name" : "validation", "num_examples" : 4795, "dataset_name" : "tatoeba_mt", "num_bytes" : 387791 }, "test" : { "name" : "test", "num_examples" : 2499, "dataset_name" : "tatoeba_mt", "num_bytes" : 202000 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 7294 }, "fra-ind" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 75212, "config_name" : "fra-ind", "splits" : { "test" : { "name" : "test", "num_bytes" : 75212, "num_examples" : 904, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 904, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "mkd-spa" : { "dataset_size" : 216, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 14994, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 14994, "dataset_name" : "tatoeba_mt", "num_examples" : 216 } }, "config_name" : "mkd-spa" }, "swe-zho" : { "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1048, "builder_name" : "tatoeba_mt", "size_in_bytes" : 68783, "config_name" : "swe-zho", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1048, "num_bytes" : 68783 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ina-spa" : { "dataset_size" : 1720, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 7849, "dataset_name" : "tatoeba_mt", "num_examples" : 108 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1612, "num_bytes" : 165605 } }, "config_name" : "ina-spa", "size_in_bytes" : 173454, "builder_name" : "tatoeba_mt" }, "lad-lfn" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 401, "builder_name" : "tatoeba_mt", "size_in_bytes" : 29203, "config_name" : "lad-lfn", "splits" : { "test" : { "num_bytes" : 23973, "num_examples" : 334, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 67, "num_bytes" : 5230, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "cat-ita" : { "supervised_keys" : null, "dataset_size" : 297, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 22469, "config_name" : "cat-ita", "splits" : { "test" : { "num_examples" : 297, "dataset_name" : "tatoeba_mt", "num_bytes" : 22469, "name" : "test" } } }, "eng-eus" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2059, "builder_name" : "tatoeba_mt", "size_in_bytes" : 163852, "config_name" : "eng-eus", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 79946, "num_examples" : 1000, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 83906, "num_examples" : 1059, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "epo-lat" : { "size_in_bytes" : 1281017, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 5979, "num_bytes" : 885465 }, "test" : { "num_bytes" : 395552, "dataset_name" : "tatoeba_mt", "num_examples" : 2799, "name" : "test" } }, "config_name" : "epo-lat", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 8778, "supervised_keys" : null }, "fin-jpn" : { "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fin-jpn", "splits" : { "test" : { "name" : "test", "num_examples" : 9933, "dataset_name" : "tatoeba_mt", "num_bytes" : 938580 }, "validation" : { "num_examples" : 10916, "dataset_name" : "tatoeba_mt", "num_bytes" : 1038239, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1976819, "supervised_keys" : null, "dataset_size" : 20849, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "heb-zho" : { "supervised_keys" : null, "dataset_size" : 708, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "heb-zho", "splits" : { "test" : { "name" : "test", "num_examples" : 708, "dataset_name" : "tatoeba_mt", "num_bytes" : 52399 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 52399 }, "lat-lfn" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 417, "config_name" : "lat-lfn", "splits" : { "test" : { "name" : "test", "num_examples" : 347, "dataset_name" : "tatoeba_mt", "num_bytes" : 22981 }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 70, "num_bytes" : 5123 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 28104, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-xal" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-xal", "splits" : { "test" : { "name" : "test", "num_bytes" : 25020, "dataset_name" : "tatoeba_mt", "num_examples" : 280 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 25020, "supervised_keys" : null, "dataset_size" : 280, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "jpn-zho" : { "config_name" : "jpn-zho", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 283939, "num_examples" : 2893, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 244984, "dataset_name" : "tatoeba_mt", "num_examples" : 2496 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 528923, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 5389 }, "jpn-nds" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 457, "supervised_keys" : null, "size_in_bytes" : 39577, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 52, "num_bytes" : 5011, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 34566, "dataset_name" : "tatoeba_mt", "num_examples" : 405 } }, "config_name" : "jpn-nds", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "por-por" : { "supervised_keys" : null, "dataset_size" : 7166, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 581539, "config_name" : "por-por", "splits" : { "validation" : { "num_bytes" : 376940, "num_examples" : 4667, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 204599, "name" : "test" } } }, "deu-cmn_Hans" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-cmn_Hans", "splits" : { "test" : { "num_bytes" : 227663, "dataset_name" : "tatoeba_mt", "num_examples" : 2608, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 4365, "num_bytes" : 385605, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 613268, "supervised_keys" : null, "dataset_size" : 6973, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-grc" : { "supervised_keys" : null, "dataset_size" : 613, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 70785, "config_name" : "eng-grc", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 613, "num_bytes" : 70785, "name" : "test" } } }, "bul-zho" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 415, "config_name" : "bul-zho", "splits" : { "test" : { "num_examples" : 415, "dataset_name" : "tatoeba_mt", "num_bytes" : 45135, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 45135, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jbo-jpn" : { "supervised_keys" : null, "dataset_size" : 920, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "jbo-jpn", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 920, "num_bytes" : 81231 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 81231 }, "jpn_Hira-cmn_Hans" : { "supervised_keys" : null, "dataset_size" : 2631, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "jpn_Hira-cmn_Hans", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1439, "num_bytes" : 148800, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 125687, "dataset_name" : "tatoeba_mt", "num_examples" : 1192 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 274487 }, "ita-nds" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 312, "config_name" : "ita-nds", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 312, "num_bytes" : 21384, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 21384, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-oci" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 54372, "config_name" : "eng-oci", "splits" : { "test" : { "num_bytes" : 54372, "num_examples" : 840, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 840, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "fin-kor_Hang" : { "supervised_keys" : null, "dataset_size" : 410, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 32896, "config_name" : "fin-kor_Hang", "splits" : { "test" : { "name" : "test", "num_examples" : 410, "dataset_name" : "tatoeba_mt", "num_bytes" : 32896 } } }, "ita-zho" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 3945, "supervised_keys" : null, "splits" : { "validation" : { "num_examples" : 1003, "dataset_name" : "tatoeba_mt", "num_bytes" : 75878, "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2942, "num_bytes" : 217098, "name" : "test" } }, "config_name" : "ita-zho", "size_in_bytes" : 292976, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "bel-rus" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 5252, "supervised_keys" : null, "size_in_bytes" : 769308, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2753, "num_bytes" : 404286, "name" : "validation" }, "test" : { "num_bytes" : 365022, "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "name" : "test" } }, "config_name" : "bel-rus", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } } }, "heb-lfn_Latn" : { "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "heb-lfn_Latn", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 40, "num_bytes" : 3422 }, "test" : { "name" : "test", "num_bytes" : 17299, "num_examples" : 211, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 20721, "supervised_keys" : null, "dataset_size" : 251, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "epo-ile" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 20798, "config_name" : "epo-ile", "splits" : { "test" : { "num_examples" : 315, "dataset_name" : "tatoeba_mt", "num_bytes" : 19439, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 22, "num_bytes" : 1359, "name" : "validation" } }, "supervised_keys" : null, "dataset_size" : 337, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "fin-swe" : { "dataset_size" : 10123, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 7283, "num_bytes" : 523109 }, "test" : { "num_examples" : 2840, "dataset_name" : "tatoeba_mt", "num_bytes" : 209037, "name" : "test" } }, "config_name" : "fin-swe", "size_in_bytes" : 732146, "builder_name" : "tatoeba_mt" }, "eng-msa" : { "supervised_keys" : null, "dataset_size" : 11891, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-msa", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 4999, "num_bytes" : 392371 }, "validation" : { "name" : "validation", "num_bytes" : 550682, "num_examples" : 6892, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 943053 }, "eng-pes" : { "dataset_size" : 3756, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 409230, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 3756, "dataset_name" : "tatoeba_mt", "num_bytes" : 409230 } }, "config_name" : "eng-pes" }, "srp_Latn-nob" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 1010915, "config_name" : "srp_Latn-nob", "splits" : { "test" : { "name" : "test", "num_examples" : 4812, "dataset_name" : "tatoeba_mt", "num_bytes" : 460044 }, "validation" : { "name" : "validation", "num_bytes" : 550871, "num_examples" : 5941, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 10753 }, "jbo-swe" : { "dataset_size" : 241, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 15034, "num_examples" : 241, "dataset_name" : "tatoeba_mt" } }, "config_name" : "jbo-swe", "size_in_bytes" : 15034, "builder_name" : "tatoeba_mt" }, "nob-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 170914, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 170914, "num_examples" : 1276, "dataset_name" : "tatoeba_mt" } }, "config_name" : "nob-rus", "dataset_size" : 1276, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "ido-ita" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 1497, "supervised_keys" : null, "size_in_bytes" : 116628, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2423, "dataset_name" : "tatoeba_mt", "num_examples" : 38 }, "test" : { "name" : "test", "num_bytes" : 114205, "num_examples" : 1459, "dataset_name" : "tatoeba_mt" } }, "config_name" : "ido-ita", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "lfn_Latn-por" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2274, "config_name" : "lfn_Latn-por", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 550, "num_bytes" : 68988, "name" : "validation" }, "test" : { "num_bytes" : 252226, "dataset_name" : "tatoeba_mt", "num_examples" : 1724, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 321214, "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "cat-deu" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 722, "config_name" : "cat-deu", "splits" : { "test" : { "num_examples" : 722, "dataset_name" : "tatoeba_mt", "num_bytes" : 59959, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 59959, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-uig_Arab" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 4024, "builder_name" : "tatoeba_mt", "size_in_bytes" : 419646, "config_name" : "eng-uig_Arab", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 105238, "dataset_name" : "tatoeba_mt", "num_examples" : 1004 }, "test" : { "num_examples" : 3020, "dataset_name" : "tatoeba_mt", "num_bytes" : 314408, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "deu-lfn" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 470, "builder_name" : "tatoeba_mt", "size_in_bytes" : 48606, "config_name" : "deu-lfn", "splits" : { "test" : { "num_examples" : 417, "dataset_name" : "tatoeba_mt", "num_bytes" : 43327, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 53, "dataset_name" : "tatoeba_mt", "num_bytes" : 5279 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "bre-eng" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 21410, "config_name" : "bre-eng", "splits" : { "test" : { "name" : "test", "num_bytes" : 21410, "dataset_name" : "tatoeba_mt", "num_examples" : 382 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 382 }, "deu-dsb" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 48185, "num_examples" : 639, "dataset_name" : "tatoeba_mt" } }, "config_name" : "deu-dsb", "size_in_bytes" : 48185, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 639, "supervised_keys" : null }, "eng-ota_Arab" : { "config_name" : "eng-ota_Arab", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 370, "num_bytes" : 33731 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 33731, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 370 }, "eng-spa" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 17632819, "config_name" : "eng-spa", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 197531, "num_bytes" : 16211897 }, "test" : { "num_examples" : 16582, "dataset_name" : "tatoeba_mt", "num_bytes" : 1420922, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 214113 }, "eng-hoc" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 659, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 36715, "num_examples" : 659, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "eng-hoc", "size_in_bytes" : 36715, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn-rus" : { "supervised_keys" : null, "dataset_size" : 25815, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 3206081, "config_name" : "jpn-rus", "splits" : { "validation" : { "num_bytes" : 1946664, "dataset_name" : "tatoeba_mt", "num_examples" : 15674, "name" : "validation" }, "test" : { "num_bytes" : 1259417, "dataset_name" : "tatoeba_mt", "num_examples" : 10141, "name" : "test" } } }, "heb-rus" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 6066, "supervised_keys" : null, "size_in_bytes" : 658874, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_examples" : 3567, "dataset_name" : "tatoeba_mt", "num_bytes" : 386010 }, "test" : { "num_bytes" : 272864, "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "name" : "test" } }, "config_name" : "heb-rus", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "deu-heb" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 4213, "config_name" : "deu-heb", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 3089, "num_bytes" : 305125 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1124, "num_bytes" : 112355, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 417480, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "tur-ukr" : { "supervised_keys" : null, "dataset_size" : 9240, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 702912, "config_name" : "tur-ukr", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 6721, "num_bytes" : 510537 }, "test" : { "name" : "test", "num_bytes" : 192375, "dataset_name" : "tatoeba_mt", "num_examples" : 2519 } } }, "jpn_Hani-rus" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 2148, "supervised_keys" : null, "size_in_bytes" : 263735, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 823, "dataset_name" : "tatoeba_mt", "num_bytes" : 100811, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1325, "num_bytes" : 162924 } }, "config_name" : "jpn_Hani-rus", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "jpn_Hira-mar" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "jpn_Hira-mar", "splits" : { "test" : { "num_bytes" : 31703, "dataset_name" : "tatoeba_mt", "num_examples" : 253, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 31703, "supervised_keys" : null, "dataset_size" : 253, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "bul-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 254650, "config_name" : "bul-rus", "splits" : { "validation" : { "num_bytes" : 112065, "num_examples" : 999, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 142585, "num_examples" : 1246, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 2245, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-lad" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 246, "builder_name" : "tatoeba_mt", "size_in_bytes" : 16527, "config_name" : "deu-lad", "splits" : { "validation" : { "num_bytes" : 2010, "num_examples" : 27, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 219, "num_bytes" : 14517, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ber_Latn-eng" : { "supervised_keys" : null, "dataset_size" : 119360, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 9789242, "config_name" : "ber_Latn-eng", "splits" : { "validation" : { "num_bytes" : 8892582, "dataset_name" : "tatoeba_mt", "num_examples" : 108399, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 896660, "dataset_name" : "tatoeba_mt", "num_examples" : 10961 } } }, "heb-ukr" : { "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "heb-ukr", "splits" : { "test" : { "num_bytes" : 83517, "dataset_name" : "tatoeba_mt", "num_examples" : 965, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 83517, "supervised_keys" : null, "dataset_size" : 965, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "nds-rus" : { "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 959, "builder_name" : "tatoeba_mt", "size_in_bytes" : 81981, "config_name" : "nds-rus", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 924, "num_bytes" : 78622, "name" : "test" }, "validation" : { "num_examples" : 35, "dataset_name" : "tatoeba_mt", "num_bytes" : 3359, "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "eng-ita" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 33971163, "config_name" : "eng-ita", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 17319, "num_bytes" : 1233733, "name" : "test" }, "validation" : { "num_bytes" : 32737430, "num_examples" : 472973, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 490292 }, "fra-ina" : { "dataset_size" : 1260, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 111106, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 84, "dataset_name" : "tatoeba_mt", "num_bytes" : 5818, "name" : "validation" }, "test" : { "num_bytes" : 105288, "num_examples" : 1176, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "fra-ina" }, "bul-epo" : { "supervised_keys" : null, "dataset_size" : 569, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "bul-epo", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 569, "num_bytes" : 56818 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 56818 }, "deu-est" : { "supervised_keys" : null, "dataset_size" : 243, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-est", "splits" : { "test" : { "name" : "test", "num_bytes" : 17015, "dataset_name" : "tatoeba_mt", "num_examples" : 243 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 17015 }, "jpn-ukr" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 476, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2025, "num_examples" : 20, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 48744, "num_examples" : 456, "dataset_name" : "tatoeba_mt" } }, "config_name" : "jpn-ukr", "size_in_bytes" : 50769, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "run-spa" : { "config_name" : "run-spa", "splits" : { "test" : { "num_bytes" : 52277, "dataset_name" : "tatoeba_mt", "num_examples" : 962, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 52277, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 962 }, "fin-hun" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fin-hun", "splits" : { "test" : { "name" : "test", "num_bytes" : 80998, "num_examples" : 1296, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 80998, "supervised_keys" : null, "dataset_size" : 1296, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "bul-ukr" : { "supervised_keys" : null, "dataset_size" : 1019, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 88243, "config_name" : "bul-ukr", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1019, "num_bytes" : 88243, "name" : "test" } } }, "deu-fin" : { "supervised_keys" : null, "dataset_size" : 9787, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 765014, "config_name" : "deu-fin", "splits" : { "validation" : { "name" : "validation", "num_examples" : 7141, "dataset_name" : "tatoeba_mt", "num_bytes" : 557059 }, "test" : { "name" : "test", "num_bytes" : 207955, "dataset_name" : "tatoeba_mt", "num_examples" : 2646 } } }, "ara-jpn_Hira" : { "config_name" : "ara-jpn_Hira", "splits" : { "test" : { "name" : "test", "num_bytes" : 111206, "num_examples" : 1091, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 111206, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1091 }, "nor-pol" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 21984, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 21984, "dataset_name" : "tatoeba_mt", "num_examples" : 280, "name" : "test" } }, "config_name" : "nor-pol", "dataset_size" : 280, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "ita-yid" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 20466, "config_name" : "ita-yid", "splits" : { "validation" : { "num_bytes" : 3609, "dataset_name" : "tatoeba_mt", "num_examples" : 36, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 205, "num_bytes" : 16857 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 241 }, "epo-srp_Cyrl" : { "supervised_keys" : null, "dataset_size" : 3820, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "epo-srp_Cyrl", "splits" : { "test" : { "num_bytes" : 203501, "num_examples" : 1794, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 229973, "num_examples" : 2026, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 433474 }, "hye-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 29846, "config_name" : "hye-rus", "splits" : { "test" : { "num_bytes" : 29846, "dataset_name" : "tatoeba_mt", "num_examples" : 226, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 226, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0" }, "bel-ukr" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 402372, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 117733, "dataset_name" : "tatoeba_mt", "num_examples" : 1020, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 2354, "dataset_name" : "tatoeba_mt", "num_bytes" : 284639 } }, "config_name" : "bel-ukr", "dataset_size" : 3374, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "tlh_Latn-cmn_Hant" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 201, "config_name" : "tlh_Latn-cmn_Hant", "splits" : { "test" : { "num_examples" : 201, "dataset_name" : "tatoeba_mt", "num_bytes" : 13574, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 13574, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ces-eng" : { "supervised_keys" : null, "dataset_size" : 30011, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 2255623, "config_name" : "ces-eng", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 16188, "num_bytes" : 1212451 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 13823, "num_bytes" : 1043172 } } }, "eng-glg" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-glg", "splits" : { "test" : { "name" : "test", "num_bytes" : 85016, "dataset_name" : "tatoeba_mt", "num_examples" : 1014 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 85016, "supervised_keys" : null, "dataset_size" : 1014, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "por-cmn_Hans" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_examples" : 637, "dataset_name" : "tatoeba_mt", "num_bytes" : 48909, "name" : "test" } }, "config_name" : "por-cmn_Hans", "size_in_bytes" : 48909, "builder_name" : "tatoeba_mt", "dataset_size" : 637, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "deu-por" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 17043, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1546200, "config_name" : "deu-por", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 7044, "num_bytes" : 651934 }, "test" : { "num_bytes" : 894266, "dataset_name" : "tatoeba_mt", "num_examples" : 9999, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "deu-nld" : { "supervised_keys" : null, "dataset_size" : 36603, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 2903363, "config_name" : "deu-nld", "splits" : { "test" : { "num_examples" : 10217, "dataset_name" : "tatoeba_mt", "num_bytes" : 811815, "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 26386, "dataset_name" : "tatoeba_mt", "num_bytes" : 2091548 } } }, "deu-vol" : { "supervised_keys" : null, "dataset_size" : 205, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 14006, "config_name" : "deu-vol", "splits" : { "test" : { "num_bytes" : 14006, "dataset_name" : "tatoeba_mt", "num_examples" : 205, "name" : "test" } } }, "deu-hrv" : { "config_name" : "deu-hrv", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 781, "num_bytes" : 58300, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 58300, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 781 }, "wuu-cmn_Hans" : { "splits" : { "validation" : { "name" : "validation", "num_examples" : 1775, "dataset_name" : "tatoeba_mt", "num_bytes" : 158748 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 872, "num_bytes" : 77430, "name" : "test" } }, "config_name" : "wuu-cmn_Hans", "size_in_bytes" : 236178, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 2647, "supervised_keys" : null }, "ido-spa" : { "size_in_bytes" : 37025, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 498, "dataset_name" : "tatoeba_mt", "num_bytes" : 37025 } }, "config_name" : "ido-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 498, "supervised_keys" : null }, "jpn-mar" : { "dataset_size" : 339, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 43170, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 43170, "dataset_name" : "tatoeba_mt", "num_examples" : 339, "name" : "test" } }, "config_name" : "jpn-mar" }, "slv-cmn_Hans" : { "supervised_keys" : null, "dataset_size" : 717, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "slv-cmn_Hans", "splits" : { "test" : { "name" : "test", "num_bytes" : 54243, "dataset_name" : "tatoeba_mt", "num_examples" : 717 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 54243 }, "tat-tur" : { "dataset_size" : 520, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 46968, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 46968, "num_examples" : 520, "dataset_name" : "tatoeba_mt" } }, "config_name" : "tat-tur" }, "bel-epo" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 75827, "config_name" : "bel-epo", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 734, "num_bytes" : 75827, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 734, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "jpn_Hira-spa" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 29882, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 9378, "num_bytes" : 938050 }, "validation" : { "num_bytes" : 2061563, "dataset_name" : "tatoeba_mt", "num_examples" : 20504, "name" : "validation" } }, "config_name" : "jpn_Hira-spa", "size_in_bytes" : 2999613, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "dtp-jpn" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 250, "builder_name" : "tatoeba_mt", "size_in_bytes" : 21300, "config_name" : "dtp-jpn", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 250, "num_bytes" : 21300 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "kzj_Latn-zsm_Latn" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 346, "supervised_keys" : null, "size_in_bytes" : 43012, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 346, "dataset_name" : "tatoeba_mt", "num_bytes" : 43012, "name" : "test" } }, "config_name" : "kzj_Latn-zsm_Latn", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "fin-nob" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "dataset_size" : 2417, "supervised_keys" : null, "splits" : { "validation" : { "num_bytes" : 49538, "dataset_name" : "tatoeba_mt", "num_examples" : 746, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 1671, "num_bytes" : 109733 } }, "config_name" : "fin-nob", "size_in_bytes" : 159271, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn_Hira-lit" : { "dataset_size" : 204, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 18266, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 204, "num_bytes" : 18266, "name" : "test" } }, "config_name" : "jpn_Hira-lit" }, "nob-nno" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 465, "config_name" : "nob-nno", "splits" : { "test" : { "num_bytes" : 31873, "dataset_name" : "tatoeba_mt", "num_examples" : 465, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 31873, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "nor-nor" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 1252, "supervised_keys" : null, "size_in_bytes" : 88270, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 271, "dataset_name" : "tatoeba_mt", "num_bytes" : 21110, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 67160, "num_examples" : 981, "dataset_name" : "tatoeba_mt" } }, "config_name" : "nor-nor", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "tur-zho" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 67341, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 985, "num_bytes" : 67341, "name" : "test" } }, "config_name" : "tur-zho", "dataset_size" : 985, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "eng-bos_Latn" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 499, "supervised_keys" : null, "splits" : { "test" : { "num_examples" : 300, "dataset_name" : "tatoeba_mt", "num_bytes" : 19082, "name" : "test" }, "validation" : { "num_bytes" : 12930, "num_examples" : 199, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "eng-bos_Latn", "size_in_bytes" : 32012, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "lat-nld" : { "dataset_size" : 365, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 24902, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 24902, "dataset_name" : "tatoeba_mt", "num_examples" : 365, "name" : "test" } }, "config_name" : "lat-nld" }, "lat-por" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 1666287, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 8303, "num_bytes" : 1040251, "name" : "validation" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 5000, "num_bytes" : 626036 } }, "config_name" : "lat-por", "dataset_size" : 13303, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 } }, "ita-ukr" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 1029315, "config_name" : "ita-ukr", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 655445, "num_examples" : 8774, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 4999, "num_bytes" : 373870 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 13773 }, "fin-tur" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 140929, "dataset_name" : "tatoeba_mt", "num_examples" : 1795 } }, "config_name" : "fin-tur", "size_in_bytes" : 140929, "builder_name" : "tatoeba_mt", "dataset_size" : 1795, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "lfn-spa" : { "size_in_bytes" : 26311, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 45, "num_bytes" : 5148 }, "test" : { "num_bytes" : 21163, "dataset_name" : "tatoeba_mt", "num_examples" : 263, "name" : "test" } }, "config_name" : "lfn-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 308, "supervised_keys" : null }, "deu-slv" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 491, "builder_name" : "tatoeba_mt", "size_in_bytes" : 30184, "config_name" : "deu-slv", "splits" : { "test" : { "name" : "test", "num_examples" : 491, "dataset_name" : "tatoeba_mt", "num_bytes" : 30184 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "eng-jpn_Hani" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_examples" : 19904, "dataset_name" : "tatoeba_mt", "num_bytes" : 1982821 }, "test" : { "num_examples" : 1499, "dataset_name" : "tatoeba_mt", "num_bytes" : 147492, "name" : "test" } }, "config_name" : "eng-jpn_Hani", "size_in_bytes" : 2130313, "builder_name" : "tatoeba_mt", "dataset_size" : 21403, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "eng-hbs" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 24221, "supervised_keys" : null, "size_in_bytes" : 2025047, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 14205, "dataset_name" : "tatoeba_mt", "num_bytes" : 1267994, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 757053, "dataset_name" : "tatoeba_mt", "num_examples" : 10016 } }, "config_name" : "eng-hbs", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "eng-tha" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 120449, "config_name" : "eng-tha", "splits" : { "test" : { "num_bytes" : 120449, "num_examples" : 1153, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1153 }, "dan-rus" : { "config_name" : "dan-rus", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 92500, "num_examples" : 1019, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 156530, "num_examples" : 1712, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 249030, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 2731 }, "epo-ron" : { "dataset_size" : 5137, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 375713, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 95906, "num_examples" : 1318, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 279807, "dataset_name" : "tatoeba_mt", "num_examples" : 3819 } }, "config_name" : "epo-ron" }, "chv-rus" : { "supervised_keys" : null, "dataset_size" : 381, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "chv-rus", "splits" : { "test" : { "num_bytes" : 40925, "num_examples" : 381, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 40925 }, "heb-yid" : { "supervised_keys" : null, "dataset_size" : 1095, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "heb-yid", "splits" : { "validation" : { "name" : "validation", "num_examples" : 195, "dataset_name" : "tatoeba_mt", "num_bytes" : 21075 }, "test" : { "num_bytes" : 98653, "dataset_name" : "tatoeba_mt", "num_examples" : 900, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 119728 }, "hrv-jpn_Hira" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 24963, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 24963, "num_examples" : 275, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "hrv-jpn_Hira", "dataset_size" : 275, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "ina-ita" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 80083, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 80083, "num_examples" : 911, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "ina-ita", "dataset_size" : 911, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "eng-tel" : { "supervised_keys" : null, "dataset_size" : 260, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-tel", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 260, "num_bytes" : 28967, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 28967 }, "cat-spa" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 131262, "dataset_name" : "tatoeba_mt", "num_examples" : 1533, "name" : "test" }, "validation" : { "num_examples" : 1293, "dataset_name" : "tatoeba_mt", "num_bytes" : 114582, "name" : "validation" } }, "config_name" : "cat-spa", "size_in_bytes" : 245844, "builder_name" : "tatoeba_mt", "dataset_size" : 2826, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "jpn-tlh" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 673, "builder_name" : "tatoeba_mt", "size_in_bytes" : 65852, "config_name" : "jpn-tlh", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 673, "num_bytes" : 65852, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "bel-lat" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 19339, "config_name" : "bel-lat", "splits" : { "test" : { "name" : "test", "num_bytes" : 19339, "num_examples" : 221, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 221 }, "bua-rus" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "bua-rus", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 805, "num_bytes" : 74438, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 74438, "supervised_keys" : null, "dataset_size" : 805, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-swe" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 4534, "supervised_keys" : null, "splits" : { "validation" : { "name" : "validation", "num_examples" : 1125, "dataset_name" : "tatoeba_mt", "num_bytes" : 84067 }, "test" : { "name" : "test", "num_bytes" : 254512, "dataset_name" : "tatoeba_mt", "num_examples" : 3409 } }, "config_name" : "deu-swe", "size_in_bytes" : 338579, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ara-epo" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 66889, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 93, "dataset_name" : "tatoeba_mt", "num_bytes" : 7818, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 59071, "dataset_name" : "tatoeba_mt", "num_examples" : 752 } }, "config_name" : "ara-epo", "dataset_size" : 845, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "eng-pol" : { "dataset_size" : 54286, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 4192770, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 3412012, "dataset_name" : "tatoeba_mt", "num_examples" : 44188 }, "test" : { "name" : "test", "num_bytes" : 780758, "dataset_name" : "tatoeba_mt", "num_examples" : 10098 } }, "config_name" : "eng-pol" }, "fra-vie" : { "config_name" : "fra-vie", "splits" : { "test" : { "name" : "test", "num_bytes" : 136035, "num_examples" : 1298, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 136035, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1298 }, "tur-uzb" : { "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "tur-uzb", "splits" : { "test" : { "num_examples" : 207, "dataset_name" : "tatoeba_mt", "num_bytes" : 14378, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 14378, "supervised_keys" : null, "dataset_size" : 207, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "heb-jpn_Hira" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 200, "supervised_keys" : null, "size_in_bytes" : 23573, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 200, "dataset_name" : "tatoeba_mt", "num_bytes" : 23573 } }, "config_name" : "heb-jpn_Hira", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } } }, "uig_Arab-cmn_Hans" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 1382, "supervised_keys" : null, "size_in_bytes" : 151405, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 1382, "dataset_name" : "tatoeba_mt", "num_bytes" : 151405 } }, "config_name" : "uig_Arab-cmn_Hans", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "pol-zho" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 1003, "supervised_keys" : null, "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1003, "num_bytes" : 82974, "name" : "test" } }, "config_name" : "pol-zho", "size_in_bytes" : 82974, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "deu-kor_Hang" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 107534, "config_name" : "deu-kor_Hang", "splits" : { "test" : { "num_bytes" : 107534, "num_examples" : 1069, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1069 }, "bul-jpn_Hira" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 33801, "num_examples" : 263, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "bul-jpn_Hira", "size_in_bytes" : 33801, "builder_name" : "tatoeba_mt", "dataset_size" : 263, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "jpn_Hira-nds" : { "dataset_size" : 395, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 49, "num_bytes" : 4705, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 29378, "num_examples" : 346, "dataset_name" : "tatoeba_mt" } }, "config_name" : "jpn_Hira-nds", "size_in_bytes" : 34083, "builder_name" : "tatoeba_mt" }, "epo-srp_Latn" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "num_bytes" : 21550, "dataset_name" : "tatoeba_mt", "num_examples" : 277, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 20921, "num_examples" : 269, "dataset_name" : "tatoeba_mt" } }, "config_name" : "epo-srp_Latn", "size_in_bytes" : 42471, "builder_name" : "tatoeba_mt", "dataset_size" : 546, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "heb-lat" : { "dataset_size" : 327, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 28580, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 4929, "num_examples" : 57, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 23651, "num_examples" : 270, "dataset_name" : "tatoeba_mt" } }, "config_name" : "heb-lat" }, "eng-tat" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 143755, "config_name" : "eng-tat", "splits" : { "test" : { "name" : "test", "num_examples" : 1450, "dataset_name" : "tatoeba_mt", "num_bytes" : 143755 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1450 }, "eng-fas" : { "supervised_keys" : null, "dataset_size" : 4791, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 522559, "config_name" : "eng-fas", "splits" : { "test" : { "num_bytes" : 410415, "num_examples" : 3761, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 1030, "dataset_name" : "tatoeba_mt", "num_bytes" : 112144 } } }, "epo-nds" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 270382, "config_name" : "epo-nds", "splits" : { "test" : { "name" : "test", "num_bytes" : 183770, "dataset_name" : "tatoeba_mt", "num_examples" : 2535 }, "validation" : { "name" : "validation", "num_bytes" : 86612, "dataset_name" : "tatoeba_mt", "num_examples" : 1180 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 3715 }, "deu-hsb" : { "supervised_keys" : null, "dataset_size" : 665, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 50779, "config_name" : "deu-hsb", "splits" : { "test" : { "num_bytes" : 50779, "num_examples" : 665, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "epo-zho" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "num_bytes" : 77989, "dataset_name" : "tatoeba_mt", "num_examples" : 1013, "name" : "validation" }, "test" : { "num_bytes" : 172567, "num_examples" : 2240, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "epo-zho", "size_in_bytes" : 250556, "builder_name" : "tatoeba_mt", "dataset_size" : 3253, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" } }, "por-toki" : { "dataset_size" : 2202, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "validation" : { "name" : "validation", "num_examples" : 484, "dataset_name" : "tatoeba_mt", "num_bytes" : 58547 }, "test" : { "num_examples" : 1718, "dataset_name" : "tatoeba_mt", "num_bytes" : 159364, "name" : "test" } }, "config_name" : "por-toki", "size_in_bytes" : 217911, "builder_name" : "tatoeba_mt" }, "fin-heb" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 211, "builder_name" : "tatoeba_mt", "size_in_bytes" : 19871, "config_name" : "fin-heb", "splits" : { "test" : { "num_bytes" : 19871, "dataset_name" : "tatoeba_mt", "num_examples" : 211, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "eng-mri" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 365, "supervised_keys" : null, "size_in_bytes" : 30911, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 30911, "dataset_name" : "tatoeba_mt", "num_examples" : 365 } }, "config_name" : "eng-mri", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } } }, "jpn-kor" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 64304, "config_name" : "jpn-kor", "splits" : { "test" : { "name" : "test", "num_examples" : 621, "dataset_name" : "tatoeba_mt", "num_bytes" : 64304 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 621 }, "rus-vie" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 34414, "config_name" : "rus-vie", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 312, "num_bytes" : 34414 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 312 }, "spa-tgl" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 630, "num_bytes" : 51801 } }, "config_name" : "spa-tgl", "size_in_bytes" : 51801, "builder_name" : "tatoeba_mt", "dataset_size" : 630, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 } }, "por-tur" : { "dataset_size" : 1793, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1793, "num_bytes" : 125270, "name" : "test" } }, "config_name" : "por-tur", "size_in_bytes" : 125270, "builder_name" : "tatoeba_mt" }, "gcf-gcf" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "size_in_bytes" : 14350, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 232, "dataset_name" : "tatoeba_mt", "num_bytes" : 14350, "name" : "test" } }, "config_name" : "gcf-gcf", "dataset_size" : 232, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "hbs-rus" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "size_in_bytes" : 634487, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 397416, "dataset_name" : "tatoeba_mt", "num_examples" : 4176 }, "test" : { "name" : "test", "num_bytes" : 237071, "dataset_name" : "tatoeba_mt", "num_examples" : 2499 } }, "config_name" : "hbs-rus", "dataset_size" : 6675, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "pol-rus" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 4556, "supervised_keys" : null, "splits" : { "test" : { "num_examples" : 3542, "dataset_name" : "tatoeba_mt", "num_bytes" : 333617, "name" : "test" }, "validation" : { "num_bytes" : 96002, "num_examples" : 1014, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "pol-rus", "size_in_bytes" : 429619, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "ita-lat" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 1735, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_examples" : 1715, "dataset_name" : "tatoeba_mt", "num_bytes" : 132924 }, "validation" : { "name" : "validation", "num_bytes" : 1314, "num_examples" : 20, "dataset_name" : "tatoeba_mt" } }, "config_name" : "ita-lat", "size_in_bytes" : 134238, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "spa-spa" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 470836, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "num_bytes" : 231019 }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2565, "num_bytes" : 239817, "name" : "validation" } }, "config_name" : "spa-spa", "dataset_size" : 5064, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 } }, "eus-spa" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2851, "builder_name" : "tatoeba_mt", "size_in_bytes" : 208039, "config_name" : "eus-spa", "splits" : { "test" : { "name" : "test", "num_bytes" : 134389, "dataset_name" : "tatoeba_mt", "num_examples" : 1849 }, "validation" : { "num_bytes" : 73650, "num_examples" : 1002, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "eng-nor" : { "supervised_keys" : null, "dataset_size" : 10705, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 863693, "config_name" : "eng-nor", "splits" : { "test" : { "num_bytes" : 405307, "num_examples" : 4999, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 458386, "num_examples" : 5706, "dataset_name" : "tatoeba_mt" } } }, "epo-lfn_Latn" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_bytes" : 60368, "num_examples" : 868, "dataset_name" : "tatoeba_mt" }, "validation" : { "name" : "validation", "num_examples" : 197, "dataset_name" : "tatoeba_mt", "num_bytes" : 15089 } }, "config_name" : "epo-lfn_Latn", "size_in_bytes" : 75457, "builder_name" : "tatoeba_mt", "dataset_size" : 1065, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "eng-tam" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 310, "supervised_keys" : null, "size_in_bytes" : 41917, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 310, "num_bytes" : 41917, "name" : "test" } }, "config_name" : "eng-tam", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "fra-pcd" : { "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fra-pcd", "splits" : { "test" : { "name" : "test", "num_examples" : 265, "dataset_name" : "tatoeba_mt", "num_bytes" : 18060 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 18060, "supervised_keys" : null, "dataset_size" : 265, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "nld-cmn_Hant" : { "config_name" : "nld-cmn_Hant", "splits" : { "test" : { "num_examples" : 1512, "dataset_name" : "tatoeba_mt", "num_bytes" : 130492, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 130492, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1512 }, "cor-eng" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 4196, "config_name" : "cor-eng", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 3197, "num_bytes" : 163747 }, "validation" : { "name" : "validation", "num_examples" : 999, "dataset_name" : "tatoeba_mt", "num_bytes" : 51376 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 215123, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-fra" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 22151012, "config_name" : "eng-fra", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 21044945, "num_examples" : 251748, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "num_bytes" : 1106067, "num_examples" : 12680, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 264428 }, "deu-kur_Latn" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 222, "supervised_keys" : null, "splits" : { "test" : { "name" : "test", "num_examples" : 222, "dataset_name" : "tatoeba_mt", "num_bytes" : 14913 } }, "config_name" : "deu-kur_Latn", "size_in_bytes" : 14913, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "epo-rus" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 57632, "config_name" : "epo-rus", "splits" : { "validation" : { "num_examples" : 45959, "dataset_name" : "tatoeba_mt", "num_bytes" : 4916871, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 1248575, "num_examples" : 11673, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 6165446, "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "epo-lad_Latn" : { "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "epo-lad_Latn", "splits" : { "test" : { "name" : "test", "num_bytes" : 17121, "num_examples" : 263, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_examples" : 30, "dataset_name" : "tatoeba_mt", "num_bytes" : 1855, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 18976, "supervised_keys" : null, "dataset_size" : 293, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-jpn" : { "dataset_size" : 41294, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 4466232, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 3232720, "num_examples" : 29867, "dataset_name" : "tatoeba_mt" }, "test" : { "num_bytes" : 1233512, "num_examples" : 11427, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "deu-jpn" }, "ita-jpn_Hira" : { "splits" : { "validation" : { "num_bytes" : 93495, "dataset_name" : "tatoeba_mt", "num_examples" : 879, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 2325, "dataset_name" : "tatoeba_mt", "num_bytes" : 241534 } }, "config_name" : "ita-jpn_Hira", "size_in_bytes" : 335029, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 3204, "supervised_keys" : null }, "hrv-ukr" : { "config_name" : "hrv-ukr", "splits" : { "test" : { "num_bytes" : 32312, "num_examples" : 388, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 32312, "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 388 }, "hbs-zho" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "size_in_bytes" : 17387, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 17387, "dataset_name" : "tatoeba_mt", "num_examples" : 235, "name" : "test" } }, "config_name" : "hbs-zho", "dataset_size" : 235, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 } }, "deu-ido" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 64799, "config_name" : "deu-ido", "splits" : { "validation" : { "num_examples" : 95, "dataset_name" : "tatoeba_mt", "num_bytes" : 5979, "name" : "validation" }, "test" : { "num_examples" : 870, "dataset_name" : "tatoeba_mt", "num_bytes" : 58820, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 965 }, "ina_Latn-tlh_Latn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 12311, "config_name" : "ina_Latn-tlh_Latn", "splits" : { "test" : { "name" : "test", "num_bytes" : 12311, "dataset_name" : "tatoeba_mt", "num_examples" : 214 } }, "supervised_keys" : null, "dataset_size" : 214, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "fra-lat" : { "size_in_bytes" : 501398, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2914, "num_bytes" : 364086, "name" : "test" }, "validation" : { "num_bytes" : 137312, "dataset_name" : "tatoeba_mt", "num_examples" : 1132, "name" : "validation" } }, "config_name" : "fra-lat", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "dataset_size" : 4046, "supervised_keys" : null }, "cym-eng" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 57737, "config_name" : "cym-eng", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 817, "num_bytes" : 57737 } }, "supervised_keys" : null, "dataset_size" : 817, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-deu" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 435915, "config_name" : "deu-deu", "splits" : { "validation" : { "num_examples" : 2114, "dataset_name" : "tatoeba_mt", "num_bytes" : 197902, "name" : "validation" }, "test" : { "num_bytes" : 238013, "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 4613, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0" }, "rus-tlh" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 19143, "config_name" : "rus-tlh", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 253, "num_bytes" : 19143 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 253 }, "ina_Latn-lad_Latn" : { "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "ina_Latn-lad_Latn", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 1848, "num_examples" : 28, "dataset_name" : "tatoeba_mt" }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 231, "num_bytes" : 16210 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 18058, "supervised_keys" : null, "dataset_size" : 259, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "nld-ron" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 248813, "config_name" : "nld-ron", "splits" : { "validation" : { "num_examples" : 1047, "dataset_name" : "tatoeba_mt", "num_bytes" : 77487, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 2268, "dataset_name" : "tatoeba_mt", "num_bytes" : 171326 } }, "supervised_keys" : null, "dataset_size" : 3315, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0" }, "eng-kaz_Cyrl" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 36935, "config_name" : "eng-kaz_Cyrl", "splits" : { "test" : { "name" : "test", "num_bytes" : 36935, "dataset_name" : "tatoeba_mt", "num_examples" : 395 } }, "supervised_keys" : null, "dataset_size" : 395, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "ina-por" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 9615, "supervised_keys" : null, "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 7079, "num_bytes" : 783848, "name" : "validation" }, "test" : { "num_bytes" : 275684, "dataset_name" : "tatoeba_mt", "num_examples" : 2536, "name" : "test" } }, "config_name" : "ina-por", "size_in_bytes" : 1059532, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fra-kor" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 32027, "config_name" : "fra-kor", "splits" : { "test" : { "name" : "test", "num_bytes" : 32027, "num_examples" : 320, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 320 }, "ina-nld" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 90977, "dataset_name" : "tatoeba_mt", "num_examples" : 1032, "name" : "test" } }, "config_name" : "ina-nld", "size_in_bytes" : 90977, "builder_name" : "tatoeba_mt", "dataset_size" : 1032, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "fas-fra" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 375, "builder_name" : "tatoeba_mt", "size_in_bytes" : 45471, "config_name" : "fas-fra", "splits" : { "test" : { "num_examples" : 375, "dataset_name" : "tatoeba_mt", "num_bytes" : 45471, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ces-lat" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 322, "config_name" : "ces-lat", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 322, "num_bytes" : 22590 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 22590, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn-vie" : { "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 584, "num_bytes" : 69734, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 122810, "dataset_name" : "tatoeba_mt", "num_examples" : 1096 } }, "config_name" : "jpn-vie", "size_in_bytes" : 192544, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 1680, "supervised_keys" : null }, "srp_Latn-ita" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 211, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 14324, "num_examples" : 211, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "srp_Latn-ita", "size_in_bytes" : 14324, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fra-jpn_Hira" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 33631, "config_name" : "fra-jpn_Hira", "splits" : { "validation" : { "name" : "validation", "num_examples" : 24841, "dataset_name" : "tatoeba_mt", "num_bytes" : 2719312 }, "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 8790, "num_bytes" : 965485 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 3684797, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "kor-rus" : { "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 220, "config_name" : "kor-rus", "splits" : { "test" : { "num_examples" : 220, "dataset_name" : "tatoeba_mt", "num_bytes" : 23083, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 23083, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "fra-tlh" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 41550, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 619, "num_bytes" : 41550, "name" : "test" } }, "config_name" : "fra-tlh", "dataset_size" : 619, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 } }, "ber_Latn-epo" : { "config_name" : "ber_Latn-epo", "splits" : { "test" : { "name" : "test", "num_bytes" : 133008, "dataset_name" : "tatoeba_mt", "num_examples" : 1574 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 133008, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1574 }, "por-spa" : { "supervised_keys" : null, "dataset_size" : 67798, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 5719346, "config_name" : "por-spa", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 56852, "num_bytes" : 4774470, "name" : "validation" }, "test" : { "num_bytes" : 944876, "num_examples" : 10946, "dataset_name" : "tatoeba_mt", "name" : "test" } } }, "kab-spa" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 70630, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 70630, "dataset_name" : "tatoeba_mt", "num_examples" : 882, "name" : "test" } }, "config_name" : "kab-spa", "dataset_size" : 882, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "ara-eng" : { "size_in_bytes" : 2716416, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_bytes" : 1778245, "dataset_name" : "tatoeba_mt", "num_examples" : 19528, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 938171, "dataset_name" : "tatoeba_mt", "num_examples" : 10304 } }, "config_name" : "ara-eng", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 29832, "supervised_keys" : null }, "eng-jbo_Latn" : { "size_in_bytes" : 873919, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "num_examples" : 6936, "dataset_name" : "tatoeba_mt", "num_bytes" : 510113, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 363806, "dataset_name" : "tatoeba_mt", "num_examples" : 4991 } }, "config_name" : "eng-jbo_Latn", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 11927, "supervised_keys" : null }, "deu-hbs" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1958, "config_name" : "deu-hbs", "splits" : { "test" : { "name" : "test", "num_bytes" : 170005, "num_examples" : 1958, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 170005, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "deu-jpn_Hani" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "deu-jpn_Hani", "splits" : { "validation" : { "num_bytes" : 355780, "dataset_name" : "tatoeba_mt", "num_examples" : 3631, "name" : "validation" }, "test" : { "num_examples" : 1289, "dataset_name" : "tatoeba_mt", "num_bytes" : 129301, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 485081, "supervised_keys" : null, "dataset_size" : 4920, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "ber-fra" : { "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 44847, "builder_name" : "tatoeba_mt", "size_in_bytes" : 3346740, "config_name" : "ber-fra", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2588616, "num_examples" : 34704, "dataset_name" : "tatoeba_mt" }, "test" : { "num_examples" : 10143, "dataset_name" : "tatoeba_mt", "num_bytes" : 758124, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "eng-slv" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 4103, "supervised_keys" : null, "size_in_bytes" : 276654, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 2494, "dataset_name" : "tatoeba_mt", "num_bytes" : 166748, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 109906, "dataset_name" : "tatoeba_mt", "num_examples" : 1609 } }, "config_name" : "eng-slv", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } } }, "eng-jav" : { "supervised_keys" : null, "dataset_size" : 261, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-jav", "splits" : { "test" : { "num_examples" : 261, "dataset_name" : "tatoeba_mt", "num_bytes" : 19245, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 19245 }, "por-tgl" : { "size_in_bytes" : 152019, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 152019, "num_examples" : 1776, "dataset_name" : "tatoeba_mt" } }, "config_name" : "por-tgl", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 1776, "supervised_keys" : null }, "spa-tur" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 28826, "builder_name" : "tatoeba_mt", "size_in_bytes" : 2006653, "config_name" : "spa-tur", "splits" : { "test" : { "num_bytes" : 743546, "num_examples" : 10614, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "num_examples" : 18212, "dataset_name" : "tatoeba_mt", "num_bytes" : 1263107 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "ido-lfn" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 27621, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 347, "dataset_name" : "tatoeba_mt", "num_bytes" : 23776, "name" : "test" }, "validation" : { "num_examples" : 54, "dataset_name" : "tatoeba_mt", "num_bytes" : 3845, "name" : "validation" } }, "config_name" : "ido-lfn", "dataset_size" : 401, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "eng-est" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "dataset_size" : 2453, "supervised_keys" : null, "size_in_bytes" : 160042, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 87767, "num_examples" : 1358, "dataset_name" : "tatoeba_mt" }, "validation" : { "name" : "validation", "num_bytes" : 72275, "dataset_name" : "tatoeba_mt", "num_examples" : 1095 } }, "config_name" : "eng-est", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } } }, "ita-vie" : { "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 249, "builder_name" : "tatoeba_mt", "size_in_bytes" : 21599, "config_name" : "ita-vie", "splits" : { "test" : { "num_examples" : 249, "dataset_name" : "tatoeba_mt", "num_bytes" : 21599, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "aze-eng" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 256850, "config_name" : "aze-eng", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2658, "num_bytes" : 185471, "name" : "test" }, "validation" : { "num_examples" : 1011, "dataset_name" : "tatoeba_mt", "num_bytes" : 71379, "name" : "validation" } }, "supervised_keys" : null, "dataset_size" : 3669, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "deu-ita" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 1921372, "config_name" : "deu-ita", "splits" : { "validation" : { "num_examples" : 12197, "dataset_name" : "tatoeba_mt", "num_bytes" : 1050058, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 871314, "num_examples" : 10093, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 22290 }, "tur-cmn_Hant" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "tur-cmn_Hant", "splits" : { "test" : { "num_examples" : 395, "dataset_name" : "tatoeba_mt", "num_bytes" : 26193, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 26193, "supervised_keys" : null, "dataset_size" : 395, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "spa-toki" : { "supervised_keys" : null, "dataset_size" : 669, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 49948, "config_name" : "spa-toki", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 30, "num_bytes" : 3933, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 639, "dataset_name" : "tatoeba_mt", "num_bytes" : 46015 } } }, "nor-swe" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "dataset_size" : 565, "supervised_keys" : null, "size_in_bytes" : 39044, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 565, "num_bytes" : 39044, "name" : "test" } }, "config_name" : "nor-swe", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "fra-gcf" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 69687, "config_name" : "fra-gcf", "splits" : { "test" : { "name" : "test", "num_bytes" : 69687, "dataset_name" : "tatoeba_mt", "num_examples" : 1163 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1163 }, "srp_Latn-rus" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 336723, "config_name" : "srp_Latn-rus", "splits" : { "test" : { "num_examples" : 1482, "dataset_name" : "tatoeba_mt", "num_bytes" : 127298, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 209425, "num_examples" : 2501, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 3983 }, "rus-uig" : { "config_name" : "rus-uig", "splits" : { "test" : { "name" : "test", "num_bytes" : 66192, "num_examples" : 534, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 66192, "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 534 }, "eng-fin" : { "config_name" : "eng-fin", "splits" : { "validation" : { "num_bytes" : 5427786, "num_examples" : 69895, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 10689, "num_bytes" : 843346, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 6271132, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 80584 }, "ota-tur" : { "size_in_bytes" : 31670, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 2304, "dataset_name" : "tatoeba_mt", "num_examples" : 23 }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 306, "num_bytes" : 29366, "name" : "test" } }, "config_name" : "ota-tur", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 329, "supervised_keys" : null }, "eng-kzj" : { "config_name" : "eng-kzj", "splits" : { "test" : { "name" : "test", "num_examples" : 1168, "dataset_name" : "tatoeba_mt", "num_bytes" : 107218 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 107218, "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 1168 }, "cor-epo" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 36954, "config_name" : "cor-epo", "splits" : { "test" : { "name" : "test", "num_bytes" : 36954, "num_examples" : 662, "dataset_name" : "tatoeba_mt" } }, "supervised_keys" : null, "dataset_size" : 662, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-hrv" : { "supervised_keys" : null, "dataset_size" : 2427, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 172219, "config_name" : "eng-hrv", "splits" : { "test" : { "num_bytes" : 104422, "dataset_name" : "tatoeba_mt", "num_examples" : 1479, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 948, "num_bytes" : 67797 } } }, "eng-vol" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 181571, "config_name" : "eng-vol", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 81012, "dataset_name" : "tatoeba_mt", "num_examples" : 1257 }, "test" : { "name" : "test", "num_examples" : 1540, "dataset_name" : "tatoeba_mt", "num_bytes" : 100559 } }, "supervised_keys" : null, "dataset_size" : 2797, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-nld" : { "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 75081, "config_name" : "eng-nld", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 62386, "num_bytes" : 4545762 }, "test" : { "num_examples" : 12695, "dataset_name" : "tatoeba_mt", "num_bytes" : 936676, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 5482438, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "kor-zho" : { "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "kor-zho", "splits" : { "test" : { "num_examples" : 410, "dataset_name" : "tatoeba_mt", "num_bytes" : 37105, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 37105, "supervised_keys" : null, "dataset_size" : 410, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0" }, "lat_Latn-por" : { "dataset_size" : 4999, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 4999, "num_bytes" : 625922, "name" : "test" } }, "config_name" : "lat_Latn-por", "size_in_bytes" : 625922, "builder_name" : "tatoeba_mt" }, "fra-uig" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 66867, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 66867, "dataset_name" : "tatoeba_mt", "num_examples" : 693 } }, "config_name" : "fra-uig", "dataset_size" : 693, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null } }, "ell-ell" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 57805, "config_name" : "ell-ell", "splits" : { "test" : { "name" : "test", "num_examples" : 532, "dataset_name" : "tatoeba_mt", "num_bytes" : 57805 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 532 }, "hun-ukr" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 472, "num_bytes" : 37775, "name" : "test" } }, "config_name" : "hun-ukr", "size_in_bytes" : 37775, "builder_name" : "tatoeba_mt", "dataset_size" : 472, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "eng-por" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 217682, "builder_name" : "tatoeba_mt", "size_in_bytes" : 16852279, "config_name" : "eng-por", "splits" : { "validation" : { "num_examples" : 204461, "dataset_name" : "tatoeba_mt", "num_bytes" : 15767111, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 1085168, "num_examples" : 13221, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "fra-ile" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 26879, "config_name" : "fra-ile", "splits" : { "test" : { "num_examples" : 393, "dataset_name" : "tatoeba_mt", "num_bytes" : 26879, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 393 }, "ell-rus" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 282508, "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "name" : "test" }, "validation" : { "num_examples" : 2531, "dataset_name" : "tatoeba_mt", "num_bytes" : 284857, "name" : "validation" } }, "config_name" : "ell-rus", "size_in_bytes" : 567365, "builder_name" : "tatoeba_mt", "dataset_size" : 5030, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "vie-zho" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 439, "builder_name" : "tatoeba_mt", "size_in_bytes" : 45707, "config_name" : "vie-zho", "splits" : { "test" : { "name" : "test", "num_examples" : 439, "dataset_name" : "tatoeba_mt", "num_bytes" : 45707 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null }, "kor_Hang-spa" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 81877, "config_name" : "kor_Hang-spa", "splits" : { "test" : { "name" : "test", "num_bytes" : 81877, "num_examples" : 920, "dataset_name" : "tatoeba_mt" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 920 }, "ces-ukr" : { "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 2899, "builder_name" : "tatoeba_mt", "size_in_bytes" : 208287, "config_name" : "ces-ukr", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1113, "num_bytes" : 81016 }, "test" : { "name" : "test", "num_bytes" : 127271, "dataset_name" : "tatoeba_mt", "num_examples" : 1786 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "bel-eng" : { "splits" : { "validation" : { "num_bytes" : 420192, "dataset_name" : "tatoeba_mt", "num_examples" : 4264, "name" : "validation" }, "test" : { "num_bytes" : 244579, "num_examples" : 2499, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "bel-eng", "size_in_bytes" : 664771, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 6763, "supervised_keys" : null }, "deu-nor" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 474094, "config_name" : "deu-nor", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 1000, "num_bytes" : 100463, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 3650, "dataset_name" : "tatoeba_mt", "num_bytes" : 373631 } }, "supervised_keys" : null, "dataset_size" : 4650, "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "nld-rus" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "nld-rus", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 6525, "num_bytes" : 613374 }, "test" : { "name" : "test", "num_examples" : 2549, "dataset_name" : "tatoeba_mt", "num_bytes" : 242444 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 855818, "supervised_keys" : null, "dataset_size" : 9074, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null }, "egl-ita" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 10674, "dataset_name" : "tatoeba_mt", "num_examples" : 201, "name" : "test" } }, "config_name" : "egl-ita", "size_in_bytes" : 10674, "builder_name" : "tatoeba_mt", "dataset_size" : 201, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "minor" : 8, "major" : 2021 } }, "ben-eng" : { "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 5146, "config_name" : "ben-eng", "splits" : { "test" : { "num_bytes" : 234586, "dataset_name" : "tatoeba_mt", "num_examples" : 2499, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 248865, "num_examples" : 2647, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 483451, "features" : { "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "jpn-lit" : { "supervised_keys" : null, "dataset_size" : 245, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "task_templates" : null, "license" : "cc-by-2.0", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "jpn-lit", "splits" : { "test" : { "name" : "test", "num_bytes" : 21568, "num_examples" : 245, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 21568 }, "afr-spa" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 29333, "config_name" : "afr-spa", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 447, "num_bytes" : 29333, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 447, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "pol-cmn_Hant" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 50414, "config_name" : "pol-cmn_Hant", "splits" : { "test" : { "name" : "test", "num_bytes" : 50414, "dataset_name" : "tatoeba_mt", "num_examples" : 595 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 595 }, "ces-epo" : { "config_name" : "ces-epo", "splits" : { "test" : { "name" : "test", "num_bytes" : 262803, "num_examples" : 3470, "dataset_name" : "tatoeba_mt" }, "validation" : { "num_bytes" : 78295, "dataset_name" : "tatoeba_mt", "num_examples" : 1057, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 341098, "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 4527 }, "rus-ukr" : { "supervised_keys" : null, "dataset_size" : 17425, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 1938831, "config_name" : "rus-ukr", "splits" : { "test" : { "name" : "test", "num_bytes" : 1107093, "dataset_name" : "tatoeba_mt", "num_examples" : 9999 }, "validation" : { "num_bytes" : 831738, "num_examples" : 7426, "dataset_name" : "tatoeba_mt", "name" : "validation" } } }, "ber-deu" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 660, "supervised_keys" : null, "size_in_bytes" : 56019, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 56019, "num_examples" : 660, "dataset_name" : "tatoeba_mt" } }, "config_name" : "ber-deu", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } } }, "fra-ukr" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 28285, "supervised_keys" : null, "size_in_bytes" : 2358605, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 10034, "num_bytes" : 836473 }, "validation" : { "name" : "validation", "num_bytes" : 1522132, "num_examples" : 18251, "dataset_name" : "tatoeba_mt" } }, "config_name" : "fra-ukr", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" }, "eng-jpn" : { "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 207925, "supervised_keys" : null, "splits" : { "test" : { "num_bytes" : 1491702, "dataset_name" : "tatoeba_mt", "num_examples" : 13861, "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 194064, "num_bytes" : 21017389, "name" : "validation" } }, "config_name" : "eng-jpn", "size_in_bytes" : 22509091, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using." }, "eng-ota_Latn" : { "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "eng-ota_Latn", "splits" : { "test" : { "num_examples" : 316, "dataset_name" : "tatoeba_mt", "num_bytes" : 24982, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 24982, "supervised_keys" : null, "dataset_size" : 316, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null }, "heb-ina" : { "size_in_bytes" : 22678, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_examples" : 216, "dataset_name" : "tatoeba_mt", "num_bytes" : 19332, "name" : "test" }, "validation" : { "num_examples" : 40, "dataset_name" : "tatoeba_mt", "num_bytes" : 3346, "name" : "validation" } }, "config_name" : "heb-ina", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "dataset_size" : 256, "supervised_keys" : null }, "eng-kur_Latn" : { "splits" : { "test" : { "num_examples" : 289, "dataset_name" : "tatoeba_mt", "num_bytes" : 18697, "name" : "test" } }, "config_name" : "eng-kur_Latn", "size_in_bytes" : 18697, "builder_name" : "tatoeba_mt", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "dataset_size" : 289, "supervised_keys" : null }, "deu-fra" : { "config_name" : "deu-fra", "splits" : { "validation" : { "num_bytes" : 8942848, "dataset_name" : "tatoeba_mt", "num_examples" : 98157, "name" : "validation" }, "test" : { "name" : "test", "num_examples" : 12417, "dataset_name" : "tatoeba_mt", "num_bytes" : 1133086 } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 10075934, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 110574 }, "bul-eng" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 1710170, "config_name" : "bul-eng", "splits" : { "validation" : { "num_bytes" : 745796, "dataset_name" : "tatoeba_mt", "num_examples" : 7799, "name" : "validation" }, "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 9999, "num_bytes" : 964374, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 17798 }, "tlh-yid" : { "features" : { "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "tlh-yid", "splits" : { "validation" : { "num_bytes" : 4091, "dataset_name" : "tatoeba_mt", "num_examples" : 48, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 23168, "num_examples" : 288, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 27259, "supervised_keys" : null, "dataset_size" : 336, "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0" }, "bul-bul" : { "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "bul-bul", "splits" : { "test" : { "name" : "test", "num_bytes" : 23925, "num_examples" : 218, "dataset_name" : "tatoeba_mt" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 23925, "supervised_keys" : null, "dataset_size" : 218, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-ido" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 258394, "config_name" : "eng-ido", "splits" : { "test" : { "num_bytes" : 149129, "dataset_name" : "tatoeba_mt", "num_examples" : 1959, "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 1482, "num_bytes" : 109265 } }, "supervised_keys" : null, "dataset_size" : 3441, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "task_templates" : null, "license" : "cc-by-2.0" }, "epo-cmn_Hant" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 90343, "config_name" : "epo-cmn_Hant", "splits" : { "test" : { "name" : "test", "num_bytes" : 60824, "num_examples" : 830, "dataset_name" : "tatoeba_mt" }, "validation" : { "name" : "validation", "num_examples" : 402, "dataset_name" : "tatoeba_mt", "num_bytes" : 29519 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1232 }, "lat-pol" : { "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 61028, "num_examples" : 914, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "lat-pol", "size_in_bytes" : 61028, "builder_name" : "tatoeba_mt", "dataset_size" : 914, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "ukr-cmn_Hans" : { "splits" : { "test" : { "name" : "test", "num_bytes" : 79448, "num_examples" : 852, "dataset_name" : "tatoeba_mt" } }, "config_name" : "ukr-cmn_Hans", "size_in_bytes" : 79448, "builder_name" : "tatoeba_mt", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "dataset_size" : 852, "supervised_keys" : null }, "eng-kor_Hang" : { "dataset_size" : 3406, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 298016, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 91437, "dataset_name" : "tatoeba_mt", "num_examples" : 1031 }, "test" : { "num_bytes" : 206579, "dataset_name" : "tatoeba_mt", "num_examples" : 2375, "name" : "test" } }, "config_name" : "eng-kor_Hang" }, "fra-jbo" : { "supervised_keys" : null, "dataset_size" : 1113, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 94434, "config_name" : "fra-jbo", "splits" : { "test" : { "name" : "test", "num_examples" : 1113, "dataset_name" : "tatoeba_mt", "num_bytes" : 94434 } } }, "eng-gsw" : { "size_in_bytes" : 9985, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 9985, "num_examples" : 204, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "eng-gsw", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "dataset_size" : 204, "supervised_keys" : null }, "hun-lat" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 43106, "config_name" : "hun-lat", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 606, "num_bytes" : 43106 } }, "supervised_keys" : null, "dataset_size" : 606, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "avk-spa" : { "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "patch" : 7, "description" : null }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 274, "builder_name" : "tatoeba_mt", "size_in_bytes" : 18491, "config_name" : "avk-spa", "splits" : { "test" : { "name" : "test", "dataset_name" : "tatoeba_mt", "num_examples" : 274, "num_bytes" : 18491 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/" }, "nld-zho" : { "dataset_size" : 1708, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "version_str" : "2021.8.7", "description" : null, "patch" : 7, "major" : 2021, "minor" : 8 }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "size_in_bytes" : 144583, "builder_name" : "tatoeba_mt", "splits" : { "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 55, "num_bytes" : 3797, "name" : "validation" }, "test" : { "name" : "test", "num_bytes" : 140786, "num_examples" : 1653, "dataset_name" : "tatoeba_mt" } }, "config_name" : "nld-zho" }, "deu-pol" : { "supervised_keys" : null, "dataset_size" : 10699, "version" : { "version_str" : "2021.8.7", "patch" : 7, "description" : null, "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 866464, "config_name" : "deu-pol", "splits" : { "validation" : { "num_examples" : 5700, "dataset_name" : "tatoeba_mt", "num_bytes" : 461017, "name" : "validation" }, "test" : { "num_bytes" : 405447, "dataset_name" : "tatoeba_mt", "num_examples" : 4999, "name" : "test" } } }, "eng-swe" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 1741388, "config_name" : "eng-swe", "splits" : { "validation" : { "name" : "validation", "num_bytes" : 1044301, "num_examples" : 15557, "dataset_name" : "tatoeba_mt" }, "test" : { "num_bytes" : 697087, "dataset_name" : "tatoeba_mt", "num_examples" : 10361, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 25918 }, "eng-mkd" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "size_in_bytes" : 6937086, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_examples" : 10009, "dataset_name" : "tatoeba_mt", "num_bytes" : 863552 }, "validation" : { "num_bytes" : 6073534, "num_examples" : 70318, "dataset_name" : "tatoeba_mt", "name" : "validation" } }, "config_name" : "eng-mkd", "dataset_size" : 80327, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" } }, "hun-kor" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 22711, "num_examples" : 270, "dataset_name" : "tatoeba_mt", "name" : "test" } }, "config_name" : "hun-kor", "size_in_bytes" : 22711, "builder_name" : "tatoeba_mt", "dataset_size" : 270, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 } }, "fra-yid" : { "dataset_size" : 443, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "num_bytes" : 35558, "num_examples" : 383, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "num_bytes" : 6092, "dataset_name" : "tatoeba_mt", "num_examples" : 60, "name" : "validation" } }, "config_name" : "fra-yid", "size_in_bytes" : 41650, "builder_name" : "tatoeba_mt" }, "spa-cmn_Hans" : { "supervised_keys" : null, "dataset_size" : 4011, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "major" : 2021, "minor" : 8 }, "license" : "cc-by-2.0", "task_templates" : null, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 296503, "config_name" : "spa-cmn_Hans", "splits" : { "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 2870, "num_bytes" : 212686 }, "test" : { "name" : "test", "num_bytes" : 83817, "num_examples" : 1141, "dataset_name" : "tatoeba_mt" } } }, "eng-tuk" : { "size_in_bytes" : 546575, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "num_bytes" : 213395, "num_examples" : 2499, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "name" : "validation", "dataset_name" : "tatoeba_mt", "num_examples" : 3865, "num_bytes" : 333180 } }, "config_name" : "eng-tuk", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "dataset_size" : 6364, "supervised_keys" : null }, "fin-spa" : { "features" : { "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "config_name" : "fin-spa", "splits" : { "test" : { "num_bytes" : 186328, "num_examples" : 2512, "dataset_name" : "tatoeba_mt", "name" : "test" }, "validation" : { "dataset_name" : "tatoeba_mt", "num_examples" : 7398, "num_bytes" : 550681, "name" : "validation" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 737009, "supervised_keys" : null, "dataset_size" : 9910, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0" }, "ita-lit" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 16502, "config_name" : "ita-lit", "splits" : { "test" : { "name" : "test", "num_examples" : 223, "dataset_name" : "tatoeba_mt", "num_bytes" : 16502 } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0", "supervised_keys" : null, "dataset_size" : 223 }, "deu-fas" : { "supervised_keys" : null, "dataset_size" : 4208, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "task_templates" : null, "license" : "cc-by-2.0", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 479819, "config_name" : "deu-fas", "splits" : { "test" : { "name" : "test", "num_bytes" : 362389, "dataset_name" : "tatoeba_mt", "num_examples" : 3184 }, "validation" : { "name" : "validation", "num_examples" : 1024, "dataset_name" : "tatoeba_mt", "num_bytes" : 117430 } } }, "deu-tat" : { "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "dataset_name" : "tatoeba_mt", "num_examples" : 2500, "num_bytes" : 371833, "name" : "test" }, "validation" : { "name" : "validation", "num_bytes" : 511413, "num_examples" : 3464, "dataset_name" : "tatoeba_mt" } }, "config_name" : "deu-tat", "size_in_bytes" : 883246, "builder_name" : "tatoeba_mt", "dataset_size" : 5964, "supervised_keys" : null, "license" : "cc-by-2.0", "task_templates" : null, "version" : { "patch" : 7, "description" : null, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 } }, "hun-jpn_Hira" : { "config_name" : "hun-jpn_Hira", "splits" : { "validation" : { "num_bytes" : 342924, "num_examples" : 3035, "dataset_name" : "tatoeba_mt", "name" : "validation" }, "test" : { "num_bytes" : 242531, "dataset_name" : "tatoeba_mt", "num_examples" : 2126, "name" : "test" } }, "builder_name" : "tatoeba_mt", "size_in_bytes" : 585455, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetlang" : { "id" : null, "_type" : "Value", "dtype" : "string" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "version" : { "major" : 2021, "minor" : 8, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 5161 }, "lat-nor" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "features" : { "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "dtype" : "string", "id" : null, "_type" : "Value" } }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "builder_name" : "tatoeba_mt", "size_in_bytes" : 23606, "config_name" : "lat-nor", "splits" : { "test" : { "num_examples" : 332, "dataset_name" : "tatoeba_mt", "num_bytes" : 23606, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 332, "version" : { "description" : null, "patch" : 7, "version_str" : "2021.8.7", "minor" : 8, "major" : 2021 }, "license" : "cc-by-2.0", "task_templates" : null }, "eng-war" : { "builder_name" : "tatoeba_mt", "size_in_bytes" : 117917, "config_name" : "eng-war", "splits" : { "test" : { "num_bytes" : 117917, "dataset_name" : "tatoeba_mt", "num_examples" : 1511, "name" : "test" } }, "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "targetString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "version" : { "major" : 2021, "minor" : 8, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "license" : "cc-by-2.0", "task_templates" : null, "supervised_keys" : null, "dataset_size" : 1511 }, "jbo-spa" : { "dataset_size" : 1506, "supervised_keys" : null, "task_templates" : null, "license" : "cc-by-2.0", "version" : { "minor" : 8, "major" : 2021, "description" : null, "patch" : 7, "version_str" : "2021.8.7" }, "post_processed" : null, "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "features" : { "sourceString" : { "dtype" : "string", "_type" : "Value", "id" : null }, "sourceLang" : { "dtype" : "string", "_type" : "Value", "id" : null }, "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "splits" : { "test" : { "name" : "test", "num_examples" : 1506, "dataset_name" : "tatoeba_mt", "num_bytes" : 107509 } }, "config_name" : "jbo-spa", "size_in_bytes" : 107509, "builder_name" : "tatoeba_mt" }, "nor-por" : { "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "features" : { "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "sourceLang" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetString" : { "_type" : "Value", "id" : null, "dtype" : "string" }, "targetlang" : { "_type" : "Value", "id" : null, "dtype" : "string" } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "builder_name" : "tatoeba_mt", "size_in_bytes" : 43546, "config_name" : "nor-por", "splits" : { "test" : { "num_bytes" : 43546, "dataset_name" : "tatoeba_mt", "num_examples" : 480, "name" : "test" } }, "supervised_keys" : null, "dataset_size" : 480, "version" : { "minor" : 8, "major" : 2021, "patch" : 7, "description" : null, "version_str" : "2021.8.7" }, "task_templates" : null, "license" : "cc-by-2.0" }, "khm-spa" : { "license" : "cc-by-2.0", "task_templates" : null, "version" : { "major" : 2021, "minor" : 8, "version_str" : "2021.8.7", "description" : null, "patch" : 7 }, "dataset_size" : 1447, "supervised_keys" : null, "size_in_bytes" : 147017, "builder_name" : "tatoeba_mt", "splits" : { "test" : { "name" : "test", "num_bytes" : 147017, "num_examples" : 1447, "dataset_name" : "tatoeba_mt" } }, "config_name" : "khm-spa", "description" : "The Tatoeba Translation Challenge is a multilingual data set of machine translation benchmarks derived from user-contributed translations collected by [Tatoeba.org](https://tatoeba.org/) and provided as parallel corpus from [OPUS](https://opus.nlpl.eu/). This dataset includes test and development data sorted by language pair. It includes test sets for hundreds of language pairs and is continuously updated. Please, check the version number tag to refer to the release that your are using.", "homepage" : "https://github.com/Helsinki-NLP/Tatoeba-Challenge/", "post_processed" : null, "features" : { "targetString" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceLang" : { "id" : null, "_type" : "Value", "dtype" : "string" }, "sourceString" : { "dtype" : "string", "id" : null, "_type" : "Value" }, "targetlang" : { "dtype" : "string", "_type" : "Value", "id" : null } }, "citation" : "@inproceedings{tiedemann-2020-tatoeba,\n title = \"The Tatoeba Translation Challenge {--} Realistic Data Sets for Low Resource and Multilingual {MT}\",\n author = {Tiedemann, J{\\\"o}rg},\n booktitle = \"Proceedings of the Fifth Conference on Machine Translation\",\n month = nov,\n year = \"2020\",\n address = \"Online\",\n publisher = \"Association for Computational Linguistics\",\n url = \"https://aclanthology.org/2020.wmt-1.139\",\n pages = \"1174--1182\",\n}" } }