Upload meta.json with huggingface_hub
Browse files
meta.json
ADDED
|
@@ -0,0 +1,103 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "3.8"
|
| 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": true,
|
| 27 |
+
"case preserved": false,
|
| 28 |
+
"description": "Russian National Corpus",
|
| 29 |
+
"id": 88,
|
| 30 |
+
"language": "rus",
|
| 31 |
+
"lemmatized": true,
|
| 32 |
+
"public": false,
|
| 33 |
+
"stop words removal": null,
|
| 34 |
+
"tagger": "UDPipe 1.2",
|
| 35 |
+
"tagset": "UPoS",
|
| 36 |
+
"tokens": 270000000,
|
| 37 |
+
"url": "http://ruscorpora.ru/"
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"NER": true,
|
| 41 |
+
"case preserved": false,
|
| 42 |
+
"description": "Russian Wikipedia dump of December 2018",
|
| 43 |
+
"id": 91,
|
| 44 |
+
"language": "rus",
|
| 45 |
+
"lemmatized": true,
|
| 46 |
+
"public": true,
|
| 47 |
+
"stop words removal": null,
|
| 48 |
+
"tagger": "UDPipe 1.2",
|
| 49 |
+
"tagset": "UPoS",
|
| 50 |
+
"tokens": 518531000,
|
| 51 |
+
"tool": "https://github.com/RaRe-Technologies/gensim/blob/master/gensim/scripts/segment_wiki.py",
|
| 52 |
+
"url": "https://dumps.wikimedia.org/"
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"NER": true,
|
| 56 |
+
"case preserved": false,
|
| 57 |
+
"description": "Russian News from Dialogue Evaluation 2020",
|
| 58 |
+
"id": 114,
|
| 59 |
+
"language": "rus",
|
| 60 |
+
"lemmatized": true,
|
| 61 |
+
"public": true,
|
| 62 |
+
"stop words removal": null,
|
| 63 |
+
"tagger": "UDPipe 1.2",
|
| 64 |
+
"tagset": "UPoS",
|
| 65 |
+
"tokens": 1321489104,
|
| 66 |
+
"url": "https://competitions.codalab.org/competitions/22168"
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"NER": true,
|
| 70 |
+
"case preserved": false,
|
| 71 |
+
"description": "Araneum Russicum Maximum",
|
| 72 |
+
"id": 115,
|
| 73 |
+
"language": "rus",
|
| 74 |
+
"lemmatized": true,
|
| 75 |
+
"public": true,
|
| 76 |
+
"stop words removal": "functional PoS",
|
| 77 |
+
"tagger": "MyStem",
|
| 78 |
+
"tagset": "UPoS",
|
| 79 |
+
"tokens": 10000000000,
|
| 80 |
+
"url": "https://rusvectores.org/en/models/"
|
| 81 |
+
}
|
| 82 |
+
],
|
| 83 |
+
"creators": [
|
| 84 |
+
{
|
| 85 |
+
"email": "[email protected]",
|
| 86 |
+
"name": "Andrey Kutuzov"
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"email": "[email protected]",
|
| 90 |
+
"name": "Maria Kunilovskaya"
|
| 91 |
+
}
|
| 92 |
+
],
|
| 93 |
+
"dimensions": 300,
|
| 94 |
+
"documentation": [
|
| 95 |
+
"https://github.com/kunilovskaya/hypohyper/"
|
| 96 |
+
],
|
| 97 |
+
"external_id": "ruscorporawikiaraneumnews_mwe_upos_cbow_300_2_2020",
|
| 98 |
+
"handle": "http://vectors.nlpl.eu/repository/20/204.zip",
|
| 99 |
+
"id": 204,
|
| 100 |
+
"iterations": 3,
|
| 101 |
+
"vocabulary size": 998459,
|
| 102 |
+
"window": 2
|
| 103 |
+
}
|