Upload folder using huggingface_hub
Browse files
examples/ytb_asr_batch1/dataset_info.json
ADDED
|
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"citation": "",
|
| 3 |
+
"description": "",
|
| 4 |
+
"features": {
|
| 5 |
+
"context": {
|
| 6 |
+
"text": {
|
| 7 |
+
"dtype": "string",
|
| 8 |
+
"_type": "Value"
|
| 9 |
+
},
|
| 10 |
+
"audio": {
|
| 11 |
+
"sampling_rate": 16000,
|
| 12 |
+
"_type": "Audio"
|
| 13 |
+
}
|
| 14 |
+
},
|
| 15 |
+
"instruction": {
|
| 16 |
+
"text": {
|
| 17 |
+
"dtype": "string",
|
| 18 |
+
"_type": "Value"
|
| 19 |
+
},
|
| 20 |
+
"audio": {
|
| 21 |
+
"sampling_rate": 16000,
|
| 22 |
+
"_type": "Audio"
|
| 23 |
+
}
|
| 24 |
+
},
|
| 25 |
+
"answer": {
|
| 26 |
+
"text": {
|
| 27 |
+
"dtype": "string",
|
| 28 |
+
"_type": "Value"
|
| 29 |
+
},
|
| 30 |
+
"audio": {
|
| 31 |
+
"sampling_rate": 16000,
|
| 32 |
+
"_type": "Audio"
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"other_attributes": {
|
| 36 |
+
"YouTube": {
|
| 37 |
+
"feature": {
|
| 38 |
+
"dtype": "string",
|
| 39 |
+
"_type": "Value"
|
| 40 |
+
},
|
| 41 |
+
"_type": "Sequence"
|
| 42 |
+
},
|
| 43 |
+
"YouTube End Time": {
|
| 44 |
+
"dtype": "float64",
|
| 45 |
+
"_type": "Value"
|
| 46 |
+
},
|
| 47 |
+
"YouTube Start Time": {
|
| 48 |
+
"dtype": "float64",
|
| 49 |
+
"_type": "Value"
|
| 50 |
+
},
|
| 51 |
+
"audio_path": {
|
| 52 |
+
"dtype": "string",
|
| 53 |
+
"_type": "Value"
|
| 54 |
+
},
|
| 55 |
+
"index": {
|
| 56 |
+
"dtype": "string",
|
| 57 |
+
"_type": "Value"
|
| 58 |
+
},
|
| 59 |
+
"segments": [
|
| 60 |
+
{
|
| 61 |
+
"end_time": {
|
| 62 |
+
"dtype": "float64",
|
| 63 |
+
"_type": "Value"
|
| 64 |
+
},
|
| 65 |
+
"start_time": {
|
| 66 |
+
"dtype": "float64",
|
| 67 |
+
"_type": "Value"
|
| 68 |
+
},
|
| 69 |
+
"text": {
|
| 70 |
+
"dtype": "string",
|
| 71 |
+
"_type": "Value"
|
| 72 |
+
}
|
| 73 |
+
}
|
| 74 |
+
]
|
| 75 |
+
},
|
| 76 |
+
"language": {
|
| 77 |
+
"dtype": "string",
|
| 78 |
+
"_type": "Value"
|
| 79 |
+
},
|
| 80 |
+
"audio_length": {
|
| 81 |
+
"dtype": "float64",
|
| 82 |
+
"_type": "Value"
|
| 83 |
+
}
|
| 84 |
+
},
|
| 85 |
+
"homepage": "",
|
| 86 |
+
"license": ""
|
| 87 |
+
}
|
examples/ytb_asr_batch1/state.json
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "7360fb5a1a8ab82b",
|
| 8 |
+
"_format_columns": [
|
| 9 |
+
"answer",
|
| 10 |
+
"audio_length",
|
| 11 |
+
"context",
|
| 12 |
+
"instruction",
|
| 13 |
+
"language",
|
| 14 |
+
"other_attributes"
|
| 15 |
+
],
|
| 16 |
+
"_format_kwargs": {},
|
| 17 |
+
"_format_type": null,
|
| 18 |
+
"_output_all_columns": false,
|
| 19 |
+
"_split": null
|
| 20 |
+
}
|
examples/ytb_asr_batch2/dataset_info.json
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"citation": "",
|
| 3 |
+
"description": "",
|
| 4 |
+
"features": {
|
| 5 |
+
"context": {
|
| 6 |
+
"text": {
|
| 7 |
+
"dtype": "string",
|
| 8 |
+
"_type": "Value"
|
| 9 |
+
},
|
| 10 |
+
"audio": {
|
| 11 |
+
"sampling_rate": 16000,
|
| 12 |
+
"_type": "Audio"
|
| 13 |
+
}
|
| 14 |
+
},
|
| 15 |
+
"instruction": {
|
| 16 |
+
"text": {
|
| 17 |
+
"dtype": "string",
|
| 18 |
+
"_type": "Value"
|
| 19 |
+
},
|
| 20 |
+
"audio": {
|
| 21 |
+
"sampling_rate": 16000,
|
| 22 |
+
"_type": "Audio"
|
| 23 |
+
}
|
| 24 |
+
},
|
| 25 |
+
"answer": {
|
| 26 |
+
"text": {
|
| 27 |
+
"dtype": "string",
|
| 28 |
+
"_type": "Value"
|
| 29 |
+
},
|
| 30 |
+
"audio": {
|
| 31 |
+
"sampling_rate": 16000,
|
| 32 |
+
"_type": "Audio"
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"other_attributes": {
|
| 36 |
+
"emotion": {
|
| 37 |
+
"dtype": "string",
|
| 38 |
+
"_type": "Value"
|
| 39 |
+
},
|
| 40 |
+
"file": {
|
| 41 |
+
"dtype": "string",
|
| 42 |
+
"_type": "Value"
|
| 43 |
+
},
|
| 44 |
+
"index": {
|
| 45 |
+
"dtype": "string",
|
| 46 |
+
"_type": "Value"
|
| 47 |
+
}
|
| 48 |
+
},
|
| 49 |
+
"language": {
|
| 50 |
+
"dtype": "string",
|
| 51 |
+
"_type": "Value"
|
| 52 |
+
},
|
| 53 |
+
"audio_length": {
|
| 54 |
+
"dtype": "float64",
|
| 55 |
+
"_type": "Value"
|
| 56 |
+
}
|
| 57 |
+
},
|
| 58 |
+
"homepage": "",
|
| 59 |
+
"license": ""
|
| 60 |
+
}
|
examples/ytb_asr_batch2/state.json
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "6c4b1fab13123a29",
|
| 8 |
+
"_format_columns": [
|
| 9 |
+
"answer",
|
| 10 |
+
"audio_length",
|
| 11 |
+
"context",
|
| 12 |
+
"instruction",
|
| 13 |
+
"language",
|
| 14 |
+
"other_attributes"
|
| 15 |
+
],
|
| 16 |
+
"_format_kwargs": {},
|
| 17 |
+
"_format_type": null,
|
| 18 |
+
"_output_all_columns": false,
|
| 19 |
+
"_split": null
|
| 20 |
+
}
|
examples/ytb_asr_batch3_malay/dataset_info.json
ADDED
|
@@ -0,0 +1,68 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"citation": "",
|
| 3 |
+
"description": "",
|
| 4 |
+
"features": {
|
| 5 |
+
"context": {
|
| 6 |
+
"text": {
|
| 7 |
+
"dtype": "string",
|
| 8 |
+
"_type": "Value"
|
| 9 |
+
},
|
| 10 |
+
"audio": {
|
| 11 |
+
"sampling_rate": 16000,
|
| 12 |
+
"_type": "Audio"
|
| 13 |
+
}
|
| 14 |
+
},
|
| 15 |
+
"instruction": {
|
| 16 |
+
"text": {
|
| 17 |
+
"dtype": "string",
|
| 18 |
+
"_type": "Value"
|
| 19 |
+
},
|
| 20 |
+
"audio": {
|
| 21 |
+
"sampling_rate": 16000,
|
| 22 |
+
"_type": "Audio"
|
| 23 |
+
}
|
| 24 |
+
},
|
| 25 |
+
"answer": {
|
| 26 |
+
"text": {
|
| 27 |
+
"dtype": "string",
|
| 28 |
+
"_type": "Value"
|
| 29 |
+
},
|
| 30 |
+
"audio": {
|
| 31 |
+
"sampling_rate": 16000,
|
| 32 |
+
"_type": "Audio"
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"language": {
|
| 36 |
+
"dtype": "string",
|
| 37 |
+
"_type": "Value"
|
| 38 |
+
},
|
| 39 |
+
"audio_length": {
|
| 40 |
+
"dtype": "float64",
|
| 41 |
+
"_type": "Value"
|
| 42 |
+
},
|
| 43 |
+
"other_attributes": {
|
| 44 |
+
"youtube_name": {
|
| 45 |
+
"dtype": "string",
|
| 46 |
+
"_type": "Value"
|
| 47 |
+
},
|
| 48 |
+
"start_time_in_sec": {
|
| 49 |
+
"dtype": "float32",
|
| 50 |
+
"_type": "Value"
|
| 51 |
+
},
|
| 52 |
+
"end_time_in_sec": {
|
| 53 |
+
"dtype": "float32",
|
| 54 |
+
"_type": "Value"
|
| 55 |
+
},
|
| 56 |
+
"link": {
|
| 57 |
+
"dtype": "string",
|
| 58 |
+
"_type": "Value"
|
| 59 |
+
},
|
| 60 |
+
"percentage_of_malay": {
|
| 61 |
+
"dtype": "string",
|
| 62 |
+
"_type": "Value"
|
| 63 |
+
}
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"homepage": "",
|
| 67 |
+
"license": ""
|
| 68 |
+
}
|
examples/ytb_asr_batch3_malay/state.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "23fffb8632becbbd",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|