seungwon12
commited on
Commit
•
ed6061b
1
Parent(s):
c06a223
Upload 3 files
Browse files- data-00000-of-00001.arrow +3 -0
- dataset_info.json +75 -0
- state.json +13 -0
data-00000-of-00001.arrow
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c31fac4982e679a665d2f319e12be4a425c396a4d8067a9b7bf21ebc4a777edf
|
3 |
+
size 315600
|
dataset_info.json
ADDED
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"builder_name": "csv",
|
3 |
+
"citation": "",
|
4 |
+
"config_name": "seungwon12--layoutlmv2_train_data",
|
5 |
+
"dataset_size": 19010,
|
6 |
+
"description": "",
|
7 |
+
"download_checksums": {
|
8 |
+
"https://huggingface.co/datasets/seungwon12/layoutlmv2_train_data/resolve/5d5c9f5b4703cc0bbdbd6acc5af70cee6c71747d/train_data%20.csv": {
|
9 |
+
"num_bytes": 18938,
|
10 |
+
"checksum": "4725296a781b5e5a0ad5c5fa35194537868b2181176b95aec11aa8d9fb127f20"
|
11 |
+
}
|
12 |
+
},
|
13 |
+
"download_size": 18938,
|
14 |
+
"features": {
|
15 |
+
"input_ids": {
|
16 |
+
"feature": {
|
17 |
+
"dtype": "int64",
|
18 |
+
"_type": "Value"
|
19 |
+
},
|
20 |
+
"_type": "Sequence"
|
21 |
+
},
|
22 |
+
"attention_mask": {
|
23 |
+
"feature": {
|
24 |
+
"dtype": "int64",
|
25 |
+
"_type": "Value"
|
26 |
+
},
|
27 |
+
"_type": "Sequence"
|
28 |
+
},
|
29 |
+
"token_type_ids": {
|
30 |
+
"feature": {
|
31 |
+
"dtype": "int64",
|
32 |
+
"_type": "Value"
|
33 |
+
},
|
34 |
+
"_type": "Sequence"
|
35 |
+
},
|
36 |
+
"bbox": {
|
37 |
+
"shape": [
|
38 |
+
512,
|
39 |
+
4
|
40 |
+
],
|
41 |
+
"dtype": "int64",
|
42 |
+
"_type": "Array2D"
|
43 |
+
},
|
44 |
+
"labels": {
|
45 |
+
"feature": {
|
46 |
+
"names": [
|
47 |
+
"header",
|
48 |
+
"key",
|
49 |
+
"value",
|
50 |
+
"table",
|
51 |
+
"other"
|
52 |
+
],
|
53 |
+
"_type": "ClassLabel"
|
54 |
+
},
|
55 |
+
"_type": "Sequence"
|
56 |
+
}
|
57 |
+
},
|
58 |
+
"homepage": "",
|
59 |
+
"license": "",
|
60 |
+
"size_in_bytes": 37948,
|
61 |
+
"splits": {
|
62 |
+
"train": {
|
63 |
+
"name": "train",
|
64 |
+
"num_bytes": 19010,
|
65 |
+
"num_examples": 9,
|
66 |
+
"dataset_name": "csv"
|
67 |
+
}
|
68 |
+
},
|
69 |
+
"version": {
|
70 |
+
"version_str": "0.0.0",
|
71 |
+
"major": 0,
|
72 |
+
"minor": 0,
|
73 |
+
"patch": 0
|
74 |
+
}
|
75 |
+
}
|
state.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_data_files": [
|
3 |
+
{
|
4 |
+
"filename": "data-00000-of-00001.arrow"
|
5 |
+
}
|
6 |
+
],
|
7 |
+
"_fingerprint": "058af2146e84995d",
|
8 |
+
"_format_columns": null,
|
9 |
+
"_format_kwargs": {},
|
10 |
+
"_format_type": "torch",
|
11 |
+
"_output_all_columns": false,
|
12 |
+
"_split": "train"
|
13 |
+
}
|