Upload meta.json with huggingface_hub
Browse files
meta.json
ADDED
|
@@ -0,0 +1,53 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": true,
|
| 28 |
+
"description": "Polish CommonCrawl Dump of December 2019",
|
| 29 |
+
"id": 117,
|
| 30 |
+
"language": "pol",
|
| 31 |
+
"lemmatized": false,
|
| 32 |
+
"public": true,
|
| 33 |
+
"stop words removal": null,
|
| 34 |
+
"tagger": null,
|
| 35 |
+
"tagset": null,
|
| 36 |
+
"tokens": 32565035188,
|
| 37 |
+
"tool": "http://dx.doi.org/10.15439/2017F432",
|
| 38 |
+
"url": "https://commoncrawl.org/"
|
| 39 |
+
}
|
| 40 |
+
],
|
| 41 |
+
"creators": [
|
| 42 |
+
{
|
| 43 |
+
"email": "[email protected]",
|
| 44 |
+
"name": "Krzysztof Wolk"
|
| 45 |
+
}
|
| 46 |
+
],
|
| 47 |
+
"dimensions": 100,
|
| 48 |
+
"handle": "http://vectors.nlpl.eu/repository/20/207.zip",
|
| 49 |
+
"id": 207,
|
| 50 |
+
"iterations": 5,
|
| 51 |
+
"vocabulary size": 35193029,
|
| 52 |
+
"window": 5
|
| 53 |
+
}
|