Upload meta.json with huggingface_hub
Browse files
meta.json
ADDED
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"algorithm": {
|
3 |
+
"command": null,
|
4 |
+
"id": 4,
|
5 |
+
"name": "Gensim Continuous Bag-of-Words",
|
6 |
+
"tool": "Gensim",
|
7 |
+
"url": "https://github.com/RaRe-Technologies/gensim",
|
8 |
+
"version": "3.6"
|
9 |
+
},
|
10 |
+
"contents": [
|
11 |
+
{
|
12 |
+
"filename": "model.txt",
|
13 |
+
"format": "text"
|
14 |
+
},
|
15 |
+
{
|
16 |
+
"filename": "model.bin",
|
17 |
+
"format": "data"
|
18 |
+
},
|
19 |
+
{
|
20 |
+
"filename": "meta.json",
|
21 |
+
"format": "json"
|
22 |
+
}
|
23 |
+
],
|
24 |
+
"corpus": [
|
25 |
+
{
|
26 |
+
"NER": false,
|
27 |
+
"case preserved": false,
|
28 |
+
"description": "Oil and Gas corpus",
|
29 |
+
"id": 75,
|
30 |
+
"language": "eng",
|
31 |
+
"lemmatized": true,
|
32 |
+
"license": "proprietary",
|
33 |
+
"public": false,
|
34 |
+
"stop words removal": "NLTK",
|
35 |
+
"tagger": "Stanford Core NLP v. 3.6.0",
|
36 |
+
"tagset": null,
|
37 |
+
"tokens": 108000000,
|
38 |
+
"tool": null,
|
39 |
+
"url": null
|
40 |
+
}
|
41 |
+
],
|
42 |
+
"creators": [
|
43 |
+
{
|
44 |
+
"email": "farhadno@ifi.uio.no",
|
45 |
+
"name": "Farhad Nooralahzadeh"
|
46 |
+
}
|
47 |
+
],
|
48 |
+
"dimensions": 400,
|
49 |
+
"handle": "http://vectors.nlpl.eu/repository/20/75.zip",
|
50 |
+
"id": 75,
|
51 |
+
"iterations": 5,
|
52 |
+
"vocabulary size": 285055,
|
53 |
+
"window": 5
|
54 |
+
}
|