{ "configuration": { "holdout_fields": [], "include_invalid": false, "ratios": { "train": 0.9, "validation": 0.1 }, "seed": 42, "stratify_fields": [ "endpoint", "language", "synthetic", "dataset" ] }, "excluded_invalid_records": 0, "holdouts": {}, "input_records": 3265, "leakage": { "audio_hash_crossings": {}, "group_crossings": {}, "is_valid": true, "metadata_key_crossings": {}, "missing_split_rows": 0 }, "records": 3265, "split_counts": { "train": 2939, "validation": 326 }, "strata": { "dataset": { "train": { "chirp3_1": 1575, "chirp3_2": 777, "chirp3_3_short": 8, "human_5": 42, "human_convcollector_1": 3, "liva_1": 326, "midcentury_1": 88, "mundo_1": 39, "orpheus_endfiller_1": 14, "orpheus_grammar_1": 14, "orpheus_midfiller_1": 14, "rime_2": 39 }, "validation": { "chirp3_1": 177, "chirp3_2": 85, "chirp3_3_short": 1, "human_5": 5, "liva_1": 36, "midcentury_1": 10, "mundo_1": 4, "orpheus_endfiller_1": 2, "orpheus_grammar_1": 1, "orpheus_midfiller_1": 1, "rime_2": 4 } }, "endpoint": { "train": { "false": 1501, "true": 1438 }, "validation": { "false": 166, "true": 160 } }, "language": { "train": { "ara": 70, "ben": 85, "dan": 83, "deu": 119, "eng": 682, "fin": 101, "fra": 145, "hin": 142, "ind": 99, "ita": 67, "jpn": 78, "kor": 89, "mar": 73, "nld": 138, "nor": 91, "pol": 91, "por": 128, "rus": 123, "spa": 185, "tur": 80, "ukr": 65, "vie": 110, "zho": 95 }, "validation": { "ara": 8, "ben": 9, "dan": 9, "deu": 13, "eng": 76, "fin": 11, "fra": 16, "hin": 16, "ind": 11, "ita": 7, "jpn": 9, "kor": 10, "mar": 8, "nld": 15, "nor": 10, "pol": 10, "por": 14, "rus": 14, "spa": 21, "tur": 9, "ukr": 7, "vie": 12, "zho": 11 } }, "synthetic": { "train": { "false": 498, "true": 2441 }, "validation": { "false": 55, "true": 271 } } } }