fc4bd1a1ec6bdffa0a7907d019e23afd3945bf4a4fceb71fc8452bb3e9d305ff

google___fleurs/yo_ng/2.0.0/af82dbec419a815084fa63ebd5d5a9f24a6e9acdf9887b9e3b8c6bbd64e0b7ac/dataset_info.json

@@ -0,0 +1 @@
+{"description": "FLEURS is the speech version of the FLORES machine translation benchmark, covering 2000 n-way parallel sentences in n=102 languages.\nFLEURS is the speech version of the FLORES machine translation benchmark, covering 2000 n-way parallel sentences in n=102 languages.", "citation": "\n", "homepage": "", "license": "", "features": {"id": {"dtype": "int32", "_type": "Value"}, "num_samples": {"dtype": "int32", "_type": "Value"}, "path": {"dtype": "string", "_type": "Value"}, "audio": {"sampling_rate": 16000, "_type": "Audio"}, "transcription": {"dtype": "string", "_type": "Value"}, "raw_transcription": {"dtype": "string", "_type": "Value"}, "gender": {"names": ["male", "female", "other"], "_type": "ClassLabel"}, "lang_id": {"names": ["af_za", "am_et", "ar_eg", "as_in", "ast_es", "az_az", "be_by", "bg_bg", "bn_in", "bs_ba", "ca_es", "ceb_ph", "ckb_iq", "cmn_hans_cn", "cs_cz", "cy_gb", "da_dk", "de_de", "el_gr", "en_us", "es_419", "et_ee", "fa_ir", "ff_sn", "fi_fi", "fil_ph", "fr_fr", "ga_ie", "gl_es", "gu_in", "ha_ng", "he_il", "hi_in", "hr_hr", "hu_hu", "hy_am", "id_id", "ig_ng", "is_is", "it_it", "ja_jp", "jv_id", "ka_ge", "kam_ke", "kea_cv", "kk_kz", "km_kh", "kn_in", "ko_kr", "ky_kg", "lb_lu", "lg_ug", "ln_cd", "lo_la", "lt_lt", "luo_ke", "lv_lv", "mi_nz", "mk_mk", "ml_in", "mn_mn", "mr_in", "ms_my", "mt_mt", "my_mm", "nb_no", "ne_np", "nl_nl", "nso_za", "ny_mw", "oc_fr", "om_et", "or_in", "pa_in", "pl_pl", "ps_af", "pt_br", "ro_ro", "ru_ru", "sd_in", "sk_sk", "sl_si", "sn_zw", "so_so", "sr_rs", "sv_se", "sw_ke", "ta_in", "te_in", "tg_tj", "th_th", "tr_tr", "uk_ua", "umb_ao", "ur_pk", "uz_uz", "vi_vn", "wo_sn", "xh_za", "yo_ng", "yue_hant_hk", "zu_za", "all"], "_type": "ClassLabel"}, "language": {"dtype": "string", "_type": "Value"}, "lang_group_id": {"names": ["western_european_we", "eastern_european_ee", "central_asia_middle_north_african_cmn", "sub_saharan_african_ssa", "south_asian_sa", "south_east_asian_sea", "chinese_japanase_korean_cjk"], "_type": "ClassLabel"}}, "supervised_keys": {"input": "audio", "output": "transcription"}, "builder_name": "fleurs", "dataset_name": "fleurs", "config_name": "yo_ng", "version": {"version_str": "2.0.0", "description": "", "major": 2, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 2312758156, "num_examples": 2339, "shard_lengths": [1000, 1000, 339], "dataset_name": "fleurs"}, "validation": {"name": "validation", "num_bytes": 394978887, "num_examples": 378, "dataset_name": "fleurs"}, "test": {"name": "test", "num_bytes": 868284634, "num_examples": 831, "dataset_name": "fleurs"}}, "download_checksums": {"data/yo_ng/audio/train.tar.gz": {"num_bytes": 1828073509, "checksum": null}, "data/yo_ng/audio/dev.tar.gz": {"num_bytes": 315960914, "checksum": null}, "data/yo_ng/audio/test.tar.gz": {"num_bytes": 692401948, "checksum": null}, "data/yo_ng/train.tsv": {"num_bytes": 1517028, "checksum": null}, "data/yo_ng/dev.tsv": {"num_bytes": 238452, "checksum": null}, "data/yo_ng/test.tsv": {"num_bytes": 535617, "checksum": null}}, "download_size": 2838727468, "dataset_size": 3576021677, "size_in_bytes": 6414749145}

google___fleurs/yo_ng/2.0.0/af82dbec419a815084fa63ebd5d5a9f24a6e9acdf9887b9e3b8c6bbd64e0b7ac/fleurs-test.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:606e6cf0ab0a96613cb2f98a5f36e196c348d483931605413dfd17bdb7dc0415
+size 868287888

google___fleurs/yo_ng/2.0.0/af82dbec419a815084fa63ebd5d5a9f24a6e9acdf9887b9e3b8c6bbd64e0b7ac/fleurs-train-00000-of-00003.arrow

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2a48d3ba8b302cb4f672f82a7d4da65d9c408c17e4e6299c3ed241361241e65f
+size 966252208