Upload folder using huggingface_hub
Browse files- examples/ytb_asr_batch3_tamil/dataset_info.json +68 -0
- examples/ytb_asr_batch3_tamil/state.json +13 -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/dataset_info.json +68 -0
- examples/ytb_sds_batch3_malay/state.json +13 -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/dataset_info.json +68 -0
- examples/ytb_sqa_batch3_chinese/state.json +13 -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/dataset_info.json +68 -0
- examples/ytb_sqa_batch3_tamil/state.json +13 -0
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/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/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/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/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/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/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 |
+
}
|