Upload folder using huggingface_hub
Browse files- examples/ytb_asr_batch3_tamil/data-00000-of-00001.arrow +3 -0
- examples/ytb_asr_batch3_tamil/dataset_info.json +68 -0
- examples/ytb_asr_batch3_tamil/state.json +13 -0
- examples/ytb_sds_batch3_chinese/data-00000-of-00001.arrow +3 -0
- examples/ytb_sds_batch3_chinese/dataset_info.json +68 -0
- examples/ytb_sds_batch3_chinese/state.json +13 -0
- examples/ytb_sds_batch3_malay/data-00000-of-00001.arrow +3 -0
- examples/ytb_sds_batch3_malay/dataset_info.json +68 -0
- examples/ytb_sds_batch3_malay/state.json +13 -0
- examples/ytb_sds_batch3_tamil/data-00000-of-00001.arrow +3 -0
- examples/ytb_sds_batch3_tamil/dataset_info.json +68 -0
- examples/ytb_sds_batch3_tamil/state.json +13 -0
- examples/ytb_sqa_batch3_chinese/data-00000-of-00001.arrow +3 -0
- examples/ytb_sqa_batch3_chinese/dataset_info.json +68 -0
- examples/ytb_sqa_batch3_chinese/state.json +13 -0
- examples/ytb_sqa_batch3_malay/data-00000-of-00001.arrow +3 -0
- examples/ytb_sqa_batch3_malay/dataset_info.json +68 -0
- examples/ytb_sqa_batch3_malay/state.json +13 -0
- examples/ytb_sqa_batch3_tamil/data-00000-of-00001.arrow +3 -0
- examples/ytb_sqa_batch3_tamil/dataset_info.json +68 -0
- examples/ytb_sqa_batch3_tamil/state.json +13 -0
examples/ytb_asr_batch3_tamil/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:610511896b4e2b7c2ade028585ae16c3d8e4f9cdbf7767e2fd1d23005f307a89
|
| 3 |
+
size 890760
|
examples/ytb_asr_batch3_tamil/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_tamil": {
|
| 61 |
+
"dtype": "string",
|
| 62 |
+
"_type": "Value"
|
| 63 |
+
}
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"homepage": "",
|
| 67 |
+
"license": ""
|
| 68 |
+
}
|
examples/ytb_asr_batch3_tamil/state.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "eb504eb1889a4aac",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|
examples/ytb_sds_batch3_chinese/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6118f3210caa8872b421495bb034b388a63cf8b14b6dc1e2e1f72347fea4e1f6
|
| 3 |
+
size 260440
|
examples/ytb_sds_batch3_chinese/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_chinese": {
|
| 61 |
+
"dtype": "string",
|
| 62 |
+
"_type": "Value"
|
| 63 |
+
}
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"homepage": "",
|
| 67 |
+
"license": ""
|
| 68 |
+
}
|
examples/ytb_sds_batch3_chinese/state.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "5525a31ce5d1f794",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|
examples/ytb_sds_batch3_malay/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d2d6b81bc2eb7c61cfa715a5fe2a67fdde7ad3db2126d2db6c10d903d30a5b9
|
| 3 |
+
size 126080
|
examples/ytb_sds_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_sds_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": "b3847ba8a7a25ab2",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|
examples/ytb_sds_batch3_tamil/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9c9d76985315c9720542b4ff8869c631e1fae751de7dce6d7f558097e8da3f85
|
| 3 |
+
size 136312
|
examples/ytb_sds_batch3_tamil/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_tamil": {
|
| 61 |
+
"dtype": "string",
|
| 62 |
+
"_type": "Value"
|
| 63 |
+
}
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"homepage": "",
|
| 67 |
+
"license": ""
|
| 68 |
+
}
|
examples/ytb_sds_batch3_tamil/state.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "4b552c5c1c6b2339",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|
examples/ytb_sqa_batch3_chinese/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d7baac023946f82a59cdacf8cb442e003433b06e59d5fba9e7999f00c981f167
|
| 3 |
+
size 165800
|
examples/ytb_sqa_batch3_chinese/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_chinese": {
|
| 61 |
+
"dtype": "string",
|
| 62 |
+
"_type": "Value"
|
| 63 |
+
}
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"homepage": "",
|
| 67 |
+
"license": ""
|
| 68 |
+
}
|
examples/ytb_sqa_batch3_chinese/state.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "13956c733fb3eeee",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|
examples/ytb_sqa_batch3_malay/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63b9d742364b5ba117a0d165cec0f838d592f0e798879203c3fb6acc6bbaa05d
|
| 3 |
+
size 126544
|
examples/ytb_sqa_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_sqa_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": "752dc89a356333ed",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|
examples/ytb_sqa_batch3_tamil/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9cf5d2182a096f441a9ca7fe35acd5a4c5925e80e9454c5782d21a0e551c3504
|
| 3 |
+
size 112272
|
examples/ytb_sqa_batch3_tamil/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_tamil": {
|
| 61 |
+
"dtype": "string",
|
| 62 |
+
"_type": "Value"
|
| 63 |
+
}
|
| 64 |
+
}
|
| 65 |
+
},
|
| 66 |
+
"homepage": "",
|
| 67 |
+
"license": ""
|
| 68 |
+
}
|
examples/ytb_sqa_batch3_tamil/state.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "4a255eac2d3b2c4d",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": null
|
| 13 |
+
}
|