| { | |
| "builder_name": "fleurs", | |
| "citation": "\n", | |
| "config_name": "ta_in", | |
| "dataset_name": "fleurs", | |
| "dataset_size": 2783252148, | |
| "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.", | |
| "download_checksums": { | |
| "data/ta_in/audio/train.tar.gz": { | |
| "num_bytes": 1638765342, | |
| "checksum": null | |
| }, | |
| "data/ta_in/audio/dev.tar.gz": { | |
| "num_bytes": 238893728, | |
| "checksum": null | |
| }, | |
| "data/ta_in/audio/test.tar.gz": { | |
| "num_bytes": 406616353, | |
| "checksum": null | |
| }, | |
| "data/ta_in/train.tsv": { | |
| "num_bytes": 3245761, | |
| "checksum": null | |
| }, | |
| "data/ta_in/dev.tsv": { | |
| "num_bytes": 507783, | |
| "checksum": null | |
| }, | |
| "data/ta_in/test.tsv": { | |
| "num_bytes": 846022, | |
| "checksum": null | |
| } | |
| }, | |
| "download_size": 2288874989, | |
| "features": { | |
| "context": { | |
| "sampling_rate": 16000, | |
| "_type": "Audio" | |
| }, | |
| "instruction": { | |
| "dtype": "string", | |
| "_type": "Value" | |
| }, | |
| "answer": { | |
| "dtype": "string", | |
| "_type": "Value" | |
| }, | |
| "audio_length": { | |
| "dtype": "float64", | |
| "_type": "Value" | |
| }, | |
| "language": { | |
| "dtype": "string", | |
| "_type": "Value" | |
| } | |
| }, | |
| "homepage": "", | |
| "license": "", | |
| "size_in_bytes": 5072127137, | |
| "splits": { | |
| "train": { | |
| "name": "train", | |
| "num_bytes": 2002058848, | |
| "num_examples": 2367, | |
| "shard_lengths": [ | |
| 1000, | |
| 1000, | |
| 367 | |
| ], | |
| "dataset_name": "fleurs" | |
| }, | |
| "validation": { | |
| "name": "validation", | |
| "num_bytes": 289505738, | |
| "num_examples": 377, | |
| "dataset_name": "fleurs" | |
| }, | |
| "test": { | |
| "name": "test", | |
| "num_bytes": 491687562, | |
| "num_examples": 591, | |
| "dataset_name": "fleurs" | |
| } | |
| }, | |
| "supervised_keys": { | |
| "input": "audio", | |
| "output": "transcription" | |
| }, | |
| "version": { | |
| "version_str": "2.0.0", | |
| "description": "", | |
| "major": 2, | |
| "minor": 0, | |
| "patch": 0 | |
| } | |
| } |