Spaces:
Sleeping
Sleeping
Upload 3 files
Browse files
tools/.datasets/data-00000-of-00001.arrow
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:36be474655d9de49931718b7d0f79f012ff519622c9aecf3e26032ae31b45b27
|
| 3 |
+
size 127640
|
tools/.datasets/dataset_info.json
ADDED
|
@@ -0,0 +1,526 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"builder_name": "gaia",
|
| 3 |
+
"citation": " ",
|
| 4 |
+
"config_name": "2023_all",
|
| 5 |
+
"dataset_name": "gaia",
|
| 6 |
+
"dataset_size": 324219,
|
| 7 |
+
"description": " ",
|
| 8 |
+
"download_checksums": {
|
| 9 |
+
"2023/test/metadata.jsonl": {
|
| 10 |
+
"num_bytes": 162773,
|
| 11 |
+
"checksum": null
|
| 12 |
+
},
|
| 13 |
+
"2023/test/e14448e9-5243-4b07-86e1-22e657f96bcf.jpg": {
|
| 14 |
+
"num_bytes": 82508,
|
| 15 |
+
"checksum": null
|
| 16 |
+
},
|
| 17 |
+
"2023/test/198ffd8f-6041-458d-bacc-fe49872cfa43.txt": {
|
| 18 |
+
"num_bytes": 378537,
|
| 19 |
+
"checksum": null
|
| 20 |
+
},
|
| 21 |
+
"2023/test/4044eab7-1282-42bd-a559-3bf3a4d5858e.pdf": {
|
| 22 |
+
"num_bytes": 64453,
|
| 23 |
+
"checksum": null
|
| 24 |
+
},
|
| 25 |
+
"2023/test/4033181f-1988-476b-bc33-6da0f96d7bd0.xlsx": {
|
| 26 |
+
"num_bytes": 5452,
|
| 27 |
+
"checksum": null
|
| 28 |
+
},
|
| 29 |
+
"2023/test/7707f3dd-1aa6-42f5-847a-b66f3eaf2ee4.png": {
|
| 30 |
+
"num_bytes": 187830,
|
| 31 |
+
"checksum": null
|
| 32 |
+
},
|
| 33 |
+
"2023/test/cd886ddd-2d12-4347-9c7a-64774f66a3d3.txt": {
|
| 34 |
+
"num_bytes": 43,
|
| 35 |
+
"checksum": null
|
| 36 |
+
},
|
| 37 |
+
"2023/test/d89733a3-7d86-4ed8-b5a3-bf4831b06e3c.jpg": {
|
| 38 |
+
"num_bytes": 6840310,
|
| 39 |
+
"checksum": null
|
| 40 |
+
},
|
| 41 |
+
"2023/test/cfd773c8-371f-425c-b081-f254f96c0530.xlsx": {
|
| 42 |
+
"num_bytes": 5623,
|
| 43 |
+
"checksum": null
|
| 44 |
+
},
|
| 45 |
+
"2023/test/ca0a4c14-4b97-43e7-8923-539d61050ae3.pdf": {
|
| 46 |
+
"num_bytes": 64453,
|
| 47 |
+
"checksum": null
|
| 48 |
+
},
|
| 49 |
+
"2023/test/c4456885-2f03-436f-8fe9-0b4ca6822cdb.pdf": {
|
| 50 |
+
"num_bytes": 64456,
|
| 51 |
+
"checksum": null
|
| 52 |
+
},
|
| 53 |
+
"2023/test/7674ee67-d671-462f-9e51-129944749a0a.png": {
|
| 54 |
+
"num_bytes": 764200,
|
| 55 |
+
"checksum": null
|
| 56 |
+
},
|
| 57 |
+
"2023/test/c7003252-fc58-44bf-92f5-ec3991a49d00.xlsx": {
|
| 58 |
+
"num_bytes": 5857,
|
| 59 |
+
"checksum": null
|
| 60 |
+
},
|
| 61 |
+
"2023/test/8f697523-6988-4c4f-8d72-760a45681f68.pdf": {
|
| 62 |
+
"num_bytes": 235914,
|
| 63 |
+
"checksum": null
|
| 64 |
+
},
|
| 65 |
+
"2023/test/a0dcc222-691e-4b03-ac75-c4493991ab80.txt": {
|
| 66 |
+
"num_bytes": 1005,
|
| 67 |
+
"checksum": null
|
| 68 |
+
},
|
| 69 |
+
"2023/test/f5d0b1c6-5e15-4c55-b60c-9fc855dda5cf.xlsx": {
|
| 70 |
+
"num_bytes": 6119,
|
| 71 |
+
"checksum": null
|
| 72 |
+
},
|
| 73 |
+
"2023/test/5b89b147-cdab-40e1-be5b-819bc076c270.mp3": {
|
| 74 |
+
"num_bytes": 17842560,
|
| 75 |
+
"checksum": null
|
| 76 |
+
},
|
| 77 |
+
"2023/test/99b5ea36-0310-49c4-85d8-9ae83a96029a.m4a": {
|
| 78 |
+
"num_bytes": 43536,
|
| 79 |
+
"checksum": null
|
| 80 |
+
},
|
| 81 |
+
"2023/test/2bb16c35-403a-4d4c-859e-a88ccd55f876.xml": {
|
| 82 |
+
"num_bytes": 143123,
|
| 83 |
+
"checksum": null
|
| 84 |
+
},
|
| 85 |
+
"2023/test/c68c0db6-1929-4194-8602-56dce5ddbd29.xml": {
|
| 86 |
+
"num_bytes": 280319,
|
| 87 |
+
"checksum": null
|
| 88 |
+
},
|
| 89 |
+
"2023/test/56376d48-f456-4c24-a917-834be04c7608.xlsx": {
|
| 90 |
+
"num_bytes": 5102,
|
| 91 |
+
"checksum": null
|
| 92 |
+
},
|
| 93 |
+
"2023/test/03c577c9-4227-48a9-9b75-f8f598de14c1.mp3": {
|
| 94 |
+
"num_bytes": 24690175,
|
| 95 |
+
"checksum": null
|
| 96 |
+
},
|
| 97 |
+
"2023/test/fe8f4748-5d00-4a27-9070-090a0cfdeac4.xlsx": {
|
| 98 |
+
"num_bytes": 5285,
|
| 99 |
+
"checksum": null
|
| 100 |
+
},
|
| 101 |
+
"2023/test/d6059b3e-e1da-43b4-ac26-ecad2984909b.csv": {
|
| 102 |
+
"num_bytes": 121076,
|
| 103 |
+
"checksum": null
|
| 104 |
+
},
|
| 105 |
+
"2023/test/943255a6-8c56-4cf8-9faf-c74743960097.csv": {
|
| 106 |
+
"num_bytes": 15114,
|
| 107 |
+
"checksum": null
|
| 108 |
+
},
|
| 109 |
+
"2023/test/b7767ed5-20c7-4243-86b1-e8bd9a3d2a64.png": {
|
| 110 |
+
"num_bytes": 1263955,
|
| 111 |
+
"checksum": null
|
| 112 |
+
},
|
| 113 |
+
"2023/test/5bbf523f-b902-4d7d-8e8d-212d00018733.mp3": {
|
| 114 |
+
"num_bytes": 10139689,
|
| 115 |
+
"checksum": null
|
| 116 |
+
},
|
| 117 |
+
"2023/test/be353748-74eb-4904-8f17-f180ce087f1a.pdf": {
|
| 118 |
+
"num_bytes": 8581099,
|
| 119 |
+
"checksum": null
|
| 120 |
+
},
|
| 121 |
+
"2023/test/70510d87-5004-4e4a-b078-21abf699dc12.txt": {
|
| 122 |
+
"num_bytes": 111168,
|
| 123 |
+
"checksum": null
|
| 124 |
+
},
|
| 125 |
+
"2023/test/7c215d46-91c7-424e-9f22-37d43ab73ea6.pdf": {
|
| 126 |
+
"num_bytes": 52620,
|
| 127 |
+
"checksum": null
|
| 128 |
+
},
|
| 129 |
+
"2023/test/86ca62df-b518-48e7-9115-1b0b800e5453.csv": {
|
| 130 |
+
"num_bytes": 316,
|
| 131 |
+
"checksum": null
|
| 132 |
+
},
|
| 133 |
+
"2023/test/7805912b-c8da-4134-9b54-b590f884352a.csv": {
|
| 134 |
+
"num_bytes": 316,
|
| 135 |
+
"checksum": null
|
| 136 |
+
},
|
| 137 |
+
"2023/test/fcd80879-4f1d-49d8-b6d6-2993607432c2.xlsx": {
|
| 138 |
+
"num_bytes": 6351,
|
| 139 |
+
"checksum": null
|
| 140 |
+
},
|
| 141 |
+
"2023/test/23bcfab0-f47b-4dcb-8599-459c329ac153.mp3": {
|
| 142 |
+
"num_bytes": 9792,
|
| 143 |
+
"checksum": null
|
| 144 |
+
},
|
| 145 |
+
"2023/test/171dd6d2-d1d4-439b-8d4e-7507018a816b.png": {
|
| 146 |
+
"num_bytes": 83202,
|
| 147 |
+
"checksum": null
|
| 148 |
+
},
|
| 149 |
+
"2023/test/680d7d77-c0c7-49c8-88fd-f8ec623645e9.pdf": {
|
| 150 |
+
"num_bytes": 64453,
|
| 151 |
+
"checksum": null
|
| 152 |
+
},
|
| 153 |
+
"2023/test/dbb02ff7-f947-491b-9ce2-41e3df16dbb8.txt": {
|
| 154 |
+
"num_bytes": 85,
|
| 155 |
+
"checksum": null
|
| 156 |
+
},
|
| 157 |
+
"2023/test/355b827f-fff0-4e0c-9ff0-65dea0609838.xlsx": {
|
| 158 |
+
"num_bytes": 5121,
|
| 159 |
+
"checksum": null
|
| 160 |
+
},
|
| 161 |
+
"2023/test/aac4df0d-407a-45f2-add5-d9b31ebe1ddc.xlsx": {
|
| 162 |
+
"num_bytes": 5623,
|
| 163 |
+
"checksum": null
|
| 164 |
+
},
|
| 165 |
+
"2023/test/8b553092-3d44-4ab3-8d1e-932aabc1e143.png": {
|
| 166 |
+
"num_bytes": 59223,
|
| 167 |
+
"checksum": null
|
| 168 |
+
},
|
| 169 |
+
"2023/test/7245af7c-404e-4d60-9ef4-94ed301e5315.jpg": {
|
| 170 |
+
"num_bytes": 82875,
|
| 171 |
+
"checksum": null
|
| 172 |
+
},
|
| 173 |
+
"2023/test/cbdb17dc-62d9-4463-b648-2eaacfeba4e5.xlsx": {
|
| 174 |
+
"num_bytes": 4773,
|
| 175 |
+
"checksum": null
|
| 176 |
+
},
|
| 177 |
+
"2023/test/07c3029f-7095-455d-a9e9-cd5e34001b38.json": {
|
| 178 |
+
"num_bytes": 280984,
|
| 179 |
+
"checksum": null
|
| 180 |
+
},
|
| 181 |
+
"2023/test/b3654e47-4307-442c-a09c-945b33b913c6.pdf": {
|
| 182 |
+
"num_bytes": 64453,
|
| 183 |
+
"checksum": null
|
| 184 |
+
},
|
| 185 |
+
"2023/test/68ccf11a-bcd3-41e5-a5ee-3e29253449e9.docx": {
|
| 186 |
+
"num_bytes": 12765,
|
| 187 |
+
"checksum": null
|
| 188 |
+
},
|
| 189 |
+
"2023/test/f1ba834a-3bcb-4e55-836c-06cc1e2ccb9f.txt": {
|
| 190 |
+
"num_bytes": 969655,
|
| 191 |
+
"checksum": null
|
| 192 |
+
},
|
| 193 |
+
"2023/test/dd024dd9-8da6-4d4e-aee1-ed0d999035a9.txt": {
|
| 194 |
+
"num_bytes": 158,
|
| 195 |
+
"checksum": null
|
| 196 |
+
},
|
| 197 |
+
"2023/test/634fca59-03b2-4cdf-9ce4-0205df22f256.pdf": {
|
| 198 |
+
"num_bytes": 109318,
|
| 199 |
+
"checksum": null
|
| 200 |
+
},
|
| 201 |
+
"2023/test/98efafc6-c376-4b53-be91-a130e1d90e02.jpg": {
|
| 202 |
+
"num_bytes": 3794,
|
| 203 |
+
"checksum": null
|
| 204 |
+
},
|
| 205 |
+
"2023/test/afd1efe6-03dd-478c-9eb1-e562355ee94e.txt": {
|
| 206 |
+
"num_bytes": 378537,
|
| 207 |
+
"checksum": null
|
| 208 |
+
},
|
| 209 |
+
"2023/test/3cc53dbf-1ab9-4d21-a56a-fc0151c10f89.xlsx": {
|
| 210 |
+
"num_bytes": 6119,
|
| 211 |
+
"checksum": null
|
| 212 |
+
},
|
| 213 |
+
"2023/test/d50b8ecb-a8aa-4696-ad84-403ef15e2c8b.pdf": {
|
| 214 |
+
"num_bytes": 53956,
|
| 215 |
+
"checksum": null
|
| 216 |
+
},
|
| 217 |
+
"2023/test/985ec22e-546b-49fc-ab3c-af490fbefdf3.txt": {
|
| 218 |
+
"num_bytes": 751161,
|
| 219 |
+
"checksum": null
|
| 220 |
+
},
|
| 221 |
+
"2023/test/5f2b2e54-5047-4394-81be-198230c3b508.xlsx": {
|
| 222 |
+
"num_bytes": 6119,
|
| 223 |
+
"checksum": null
|
| 224 |
+
},
|
| 225 |
+
"2023/test/59b3cfb0-a06a-4ac5-b54e-81c9db8b0957.xlsx": {
|
| 226 |
+
"num_bytes": 5857,
|
| 227 |
+
"checksum": null
|
| 228 |
+
},
|
| 229 |
+
"2023/test/32f386b9-73ee-4455-b412-ddad508aa979.pdf": {
|
| 230 |
+
"num_bytes": 52620,
|
| 231 |
+
"checksum": null
|
| 232 |
+
},
|
| 233 |
+
"2023/test/d62cbee6-47c7-4918-825d-3b73b1af7e85.png": {
|
| 234 |
+
"num_bytes": 77028,
|
| 235 |
+
"checksum": null
|
| 236 |
+
},
|
| 237 |
+
"2023/test/91f2bf12-5280-4efc-b9a7-26e67ca850b4.jpg": {
|
| 238 |
+
"num_bytes": 96084,
|
| 239 |
+
"checksum": null
|
| 240 |
+
},
|
| 241 |
+
"2023/test/063800f6-8832-4856-972b-17b877612533.png": {
|
| 242 |
+
"num_bytes": 13859,
|
| 243 |
+
"checksum": null
|
| 244 |
+
},
|
| 245 |
+
"2023/test/dfa03d6c-402b-43fc-9222-5738f8bdfd0c.txt": {
|
| 246 |
+
"num_bytes": 11609,
|
| 247 |
+
"checksum": null
|
| 248 |
+
},
|
| 249 |
+
"2023/test/e51753c7-3ef3-4404-a352-11a18e5760c9.png": {
|
| 250 |
+
"num_bytes": 502276,
|
| 251 |
+
"checksum": null
|
| 252 |
+
},
|
| 253 |
+
"2023/test/7a770333-8c1b-4008-b630-9d3cb4f0c171.txt": {
|
| 254 |
+
"num_bytes": 660753,
|
| 255 |
+
"checksum": null
|
| 256 |
+
},
|
| 257 |
+
"2023/test/52e8ce1c-09bd-4537-8e2d-67d1648779b9.csv": {
|
| 258 |
+
"num_bytes": 316,
|
| 259 |
+
"checksum": null
|
| 260 |
+
},
|
| 261 |
+
"2023/test/b74b4ce7-4f03-42b5-b60e-62da7ffa282e.xlsx": {
|
| 262 |
+
"num_bytes": 6171,
|
| 263 |
+
"checksum": null
|
| 264 |
+
},
|
| 265 |
+
"2023/test/d366cc70-86f1-4dca-bf12-440479c825fe.pptx": {
|
| 266 |
+
"num_bytes": 98790,
|
| 267 |
+
"checksum": null
|
| 268 |
+
},
|
| 269 |
+
"2023/test/021a5339-744f-42b7-bd9b-9368b3efda7a.pdf": {
|
| 270 |
+
"num_bytes": 64453,
|
| 271 |
+
"checksum": null
|
| 272 |
+
},
|
| 273 |
+
"2023/test/aea1ea38-dfd0-41ab-ad79-badc3c69c784.txt": {
|
| 274 |
+
"num_bytes": 32,
|
| 275 |
+
"checksum": null
|
| 276 |
+
},
|
| 277 |
+
"2023/test/0c393561-dd13-4b7c-ac49-20ac469aa276.MOV": {
|
| 278 |
+
"num_bytes": 23012229,
|
| 279 |
+
"checksum": null
|
| 280 |
+
},
|
| 281 |
+
"2023/test/9ca9bc20-1a0b-4076-9937-7724e3491cf8.png": {
|
| 282 |
+
"num_bytes": 101283,
|
| 283 |
+
"checksum": null
|
| 284 |
+
},
|
| 285 |
+
"2023/test/d8434132-f196-4048-82c3-c06facff53c0.xlsx": {
|
| 286 |
+
"num_bytes": 5452,
|
| 287 |
+
"checksum": null
|
| 288 |
+
},
|
| 289 |
+
"2023/test/f6d29ef1-0e4d-41cb-ac25-e60023b3bd96.xlsx": {
|
| 290 |
+
"num_bytes": 6171,
|
| 291 |
+
"checksum": null
|
| 292 |
+
},
|
| 293 |
+
"2023/test/4cf4a5c1-7c9c-4cce-94cb-57b8be196244.png": {
|
| 294 |
+
"num_bytes": 13263,
|
| 295 |
+
"checksum": null
|
| 296 |
+
},
|
| 297 |
+
"2023/validation/metadata.jsonl": {
|
| 298 |
+
"num_bytes": 222894,
|
| 299 |
+
"checksum": null
|
| 300 |
+
},
|
| 301 |
+
"2023/validation/32102e3e-d12a-4209-9163-7b3a104efe5d.xlsx": {
|
| 302 |
+
"num_bytes": 6119,
|
| 303 |
+
"checksum": null
|
| 304 |
+
},
|
| 305 |
+
"2023/validation/7dd30055-0198-452e-8c25-f73dbe27dcb8.pdb": {
|
| 306 |
+
"num_bytes": 2897289,
|
| 307 |
+
"checksum": null
|
| 308 |
+
},
|
| 309 |
+
"2023/validation/bec74516-02fc-48dc-b202-55e78d0e17cf.jsonld": {
|
| 310 |
+
"num_bytes": 3904,
|
| 311 |
+
"checksum": null
|
| 312 |
+
},
|
| 313 |
+
"2023/validation/df6561b2-7ee5-4540-baab-5095f742716a.png": {
|
| 314 |
+
"num_bytes": 16447,
|
| 315 |
+
"checksum": null
|
| 316 |
+
},
|
| 317 |
+
"2023/validation/cffe0e32-c9a6-4c52-9877-78ceb4aaa9fb.docx": {
|
| 318 |
+
"num_bytes": 17525,
|
| 319 |
+
"checksum": null
|
| 320 |
+
},
|
| 321 |
+
"2023/validation/5cfb274c-0207-4aa7-9575-6ac0bd95d9b2.xlsx": {
|
| 322 |
+
"num_bytes": 5115,
|
| 323 |
+
"checksum": null
|
| 324 |
+
},
|
| 325 |
+
"2023/validation/9b54f9d9-35ee-4a14-b62f-d130ea00317f.zip": {
|
| 326 |
+
"num_bytes": 11689,
|
| 327 |
+
"checksum": null
|
| 328 |
+
},
|
| 329 |
+
"2023/validation/2b3ef98c-cc05-450b-a719-711aee40ac65.mp3": {
|
| 330 |
+
"num_bytes": 205008,
|
| 331 |
+
"checksum": null
|
| 332 |
+
},
|
| 333 |
+
"2023/validation/bfcd99e1-0690-4b53-a85c-0174a8629083.zip": {
|
| 334 |
+
"num_bytes": 162769,
|
| 335 |
+
"checksum": null
|
| 336 |
+
},
|
| 337 |
+
"2023/validation/076c8171-9b3b-49b9-a477-244d2a532826.xlsx": {
|
| 338 |
+
"num_bytes": 6171,
|
| 339 |
+
"checksum": null
|
| 340 |
+
},
|
| 341 |
+
"2023/validation/cca530fc-4052-43b2-b130-b30968d8aa44.png": {
|
| 342 |
+
"num_bytes": 63080,
|
| 343 |
+
"checksum": null
|
| 344 |
+
},
|
| 345 |
+
"2023/validation/e9a2c537-8232-4c3f-85b0-b52de6bcba99.pdf": {
|
| 346 |
+
"num_bytes": 64453,
|
| 347 |
+
"checksum": null
|
| 348 |
+
},
|
| 349 |
+
"2023/validation/9318445f-fe6a-4e1b-acbf-c68228c9906a.png": {
|
| 350 |
+
"num_bytes": 133568,
|
| 351 |
+
"checksum": null
|
| 352 |
+
},
|
| 353 |
+
"2023/validation/8f80e01c-1296-4371-9486-bb3d68651a60.png": {
|
| 354 |
+
"num_bytes": 560,
|
| 355 |
+
"checksum": null
|
| 356 |
+
},
|
| 357 |
+
"2023/validation/366e2f2b-8632-4ef2-81eb-bc3877489217.pdf": {
|
| 358 |
+
"num_bytes": 53956,
|
| 359 |
+
"checksum": null
|
| 360 |
+
},
|
| 361 |
+
"2023/validation/c526d8d6-5987-4da9-b24c-83466fa172f3.xlsx": {
|
| 362 |
+
"num_bytes": 12235,
|
| 363 |
+
"checksum": null
|
| 364 |
+
},
|
| 365 |
+
"2023/validation/389793a7-ca17-4e82-81cb-2b3a2391b4b9.txt": {
|
| 366 |
+
"num_bytes": 99,
|
| 367 |
+
"checksum": null
|
| 368 |
+
},
|
| 369 |
+
"2023/validation/3da89939-209c-4086-8520-7eb734e6b4ef.xlsx": {
|
| 370 |
+
"num_bytes": 11898,
|
| 371 |
+
"checksum": null
|
| 372 |
+
},
|
| 373 |
+
"2023/validation/a3fbeb63-0e8c-4a11-bff6-0e3b484c3e9c.pptx": {
|
| 374 |
+
"num_bytes": 388996,
|
| 375 |
+
"checksum": null
|
| 376 |
+
},
|
| 377 |
+
"2023/validation/8d46b8d6-b38a-47ff-ac74-cda14cf2d19b.csv": {
|
| 378 |
+
"num_bytes": 13478,
|
| 379 |
+
"checksum": null
|
| 380 |
+
},
|
| 381 |
+
"2023/validation/54612da3-fd56-4941-80f4-5eb82330de25.xlsx": {
|
| 382 |
+
"num_bytes": 5623,
|
| 383 |
+
"checksum": null
|
| 384 |
+
},
|
| 385 |
+
"2023/validation/6359a0b1-8f7b-499b-9336-840f9ab90688.png": {
|
| 386 |
+
"num_bytes": 6161,
|
| 387 |
+
"checksum": null
|
| 388 |
+
},
|
| 389 |
+
"2023/validation/7cc4acfa-63fd-4acc-a1a1-e8e529e0a97f.xlsx": {
|
| 390 |
+
"num_bytes": 5285,
|
| 391 |
+
"checksum": null
|
| 392 |
+
},
|
| 393 |
+
"2023/validation/65afbc8a-89ca-4ad5-8d62-355bb401f61d.xlsx": {
|
| 394 |
+
"num_bytes": 12370,
|
| 395 |
+
"checksum": null
|
| 396 |
+
},
|
| 397 |
+
"2023/validation/cca70ce6-1952-45d2-acd4-80c903b0bc49.png": {
|
| 398 |
+
"num_bytes": 37639,
|
| 399 |
+
"checksum": null
|
| 400 |
+
},
|
| 401 |
+
"2023/validation/99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3.mp3": {
|
| 402 |
+
"num_bytes": 179304,
|
| 403 |
+
"checksum": null
|
| 404 |
+
},
|
| 405 |
+
"2023/validation/b7f857e4-d8aa-4387-af2a-0e844df5b9d8.png": {
|
| 406 |
+
"num_bytes": 23217,
|
| 407 |
+
"checksum": null
|
| 408 |
+
},
|
| 409 |
+
"2023/validation/d8152ad6-e4d5-4c12-8bb7-8d57dc10c6de.png": {
|
| 410 |
+
"num_bytes": 21240,
|
| 411 |
+
"checksum": null
|
| 412 |
+
},
|
| 413 |
+
"2023/validation/67e8878b-5cef-4375-804e-e6291fdbe78a.pdf": {
|
| 414 |
+
"num_bytes": 53956,
|
| 415 |
+
"checksum": null
|
| 416 |
+
},
|
| 417 |
+
"2023/validation/4d51c4bf-4b0e-4f3d-897b-3f6687a7d9f2.xlsx": {
|
| 418 |
+
"num_bytes": 5857,
|
| 419 |
+
"checksum": null
|
| 420 |
+
},
|
| 421 |
+
"2023/validation/f918266a-b3e0-4914-865d-4faa564f1aef.py": {
|
| 422 |
+
"num_bytes": 698,
|
| 423 |
+
"checksum": null
|
| 424 |
+
},
|
| 425 |
+
"2023/validation/da52d699-e8d2-4dc5-9191-a2199e0b6a9b.xlsx": {
|
| 426 |
+
"num_bytes": 5452,
|
| 427 |
+
"checksum": null
|
| 428 |
+
},
|
| 429 |
+
"2023/validation/5b2a14e8-6e59-479c-80e3-4696e8980152.jpg": {
|
| 430 |
+
"num_bytes": 1738521,
|
| 431 |
+
"checksum": null
|
| 432 |
+
},
|
| 433 |
+
"2023/validation/edd4d4f2-1a58-45c4-b038-67337af4e029.xlsx": {
|
| 434 |
+
"num_bytes": 5623,
|
| 435 |
+
"checksum": null
|
| 436 |
+
},
|
| 437 |
+
"2023/validation/4d0aa727-86b1-406b-9b33-f870dd14a4a5.xlsx": {
|
| 438 |
+
"num_bytes": 5623,
|
| 439 |
+
"checksum": null
|
| 440 |
+
},
|
| 441 |
+
"2023/validation/1f975693-876d-457b-a649-393859e79bf3.mp3": {
|
| 442 |
+
"num_bytes": 280868,
|
| 443 |
+
"checksum": null
|
| 444 |
+
},
|
| 445 |
+
"2023/validation/b2c257e0-3ad7-4f05-b8e3-d9da973be36e.jpg": {
|
| 446 |
+
"num_bytes": 3597860,
|
| 447 |
+
"checksum": null
|
| 448 |
+
},
|
| 449 |
+
"2023/validation/7bd855d8-463d-4ed5-93ca-5fe35145f733.xlsx": {
|
| 450 |
+
"num_bytes": 5285,
|
| 451 |
+
"checksum": null
|
| 452 |
+
}
|
| 453 |
+
},
|
| 454 |
+
"download_size": 110139584,
|
| 455 |
+
"features": {
|
| 456 |
+
"task_id": {
|
| 457 |
+
"dtype": "string",
|
| 458 |
+
"_type": "Value"
|
| 459 |
+
},
|
| 460 |
+
"Question": {
|
| 461 |
+
"dtype": "string",
|
| 462 |
+
"_type": "Value"
|
| 463 |
+
},
|
| 464 |
+
"Level": {
|
| 465 |
+
"dtype": "string",
|
| 466 |
+
"_type": "Value"
|
| 467 |
+
},
|
| 468 |
+
"Final answer": {
|
| 469 |
+
"dtype": "string",
|
| 470 |
+
"_type": "Value"
|
| 471 |
+
},
|
| 472 |
+
"file_name": {
|
| 473 |
+
"dtype": "string",
|
| 474 |
+
"_type": "Value"
|
| 475 |
+
},
|
| 476 |
+
"file_path": {
|
| 477 |
+
"dtype": "string",
|
| 478 |
+
"_type": "Value"
|
| 479 |
+
},
|
| 480 |
+
"Annotator Metadata": {
|
| 481 |
+
"Steps": {
|
| 482 |
+
"dtype": "string",
|
| 483 |
+
"_type": "Value"
|
| 484 |
+
},
|
| 485 |
+
"Number of steps": {
|
| 486 |
+
"dtype": "string",
|
| 487 |
+
"_type": "Value"
|
| 488 |
+
},
|
| 489 |
+
"How long did this take?": {
|
| 490 |
+
"dtype": "string",
|
| 491 |
+
"_type": "Value"
|
| 492 |
+
},
|
| 493 |
+
"Tools": {
|
| 494 |
+
"dtype": "string",
|
| 495 |
+
"_type": "Value"
|
| 496 |
+
},
|
| 497 |
+
"Number of tools": {
|
| 498 |
+
"dtype": "string",
|
| 499 |
+
"_type": "Value"
|
| 500 |
+
}
|
| 501 |
+
}
|
| 502 |
+
},
|
| 503 |
+
"homepage": "",
|
| 504 |
+
"license": "",
|
| 505 |
+
"size_in_bytes": 110463803,
|
| 506 |
+
"splits": {
|
| 507 |
+
"test": {
|
| 508 |
+
"name": "test",
|
| 509 |
+
"num_bytes": 125374,
|
| 510 |
+
"num_examples": 301,
|
| 511 |
+
"dataset_name": "gaia"
|
| 512 |
+
},
|
| 513 |
+
"validation": {
|
| 514 |
+
"name": "validation",
|
| 515 |
+
"num_bytes": 198845,
|
| 516 |
+
"num_examples": 165,
|
| 517 |
+
"dataset_name": "gaia"
|
| 518 |
+
}
|
| 519 |
+
},
|
| 520 |
+
"version": {
|
| 521 |
+
"version_str": "0.0.1",
|
| 522 |
+
"major": 0,
|
| 523 |
+
"minor": 0,
|
| 524 |
+
"patch": 1
|
| 525 |
+
}
|
| 526 |
+
}
|
tools/.datasets/state.json
ADDED
|
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_data_files": [
|
| 3 |
+
{
|
| 4 |
+
"filename": "data-00000-of-00001.arrow"
|
| 5 |
+
}
|
| 6 |
+
],
|
| 7 |
+
"_fingerprint": "d6d288025e932213",
|
| 8 |
+
"_format_columns": null,
|
| 9 |
+
"_format_kwargs": {},
|
| 10 |
+
"_format_type": null,
|
| 11 |
+
"_output_all_columns": false,
|
| 12 |
+
"_split": "test"
|
| 13 |
+
}
|