| { | |
| "algorithm": { | |
| "command": null, | |
| "id": 4, | |
| "name": "Gensim Continuous Bag-of-Words", | |
| "tool": "Gensim", | |
| "url": "https://github.com/RaRe-Technologies/gensim", | |
| "version": "3.6" | |
| }, | |
| "contents": [ | |
| { | |
| "filename": "model.txt", | |
| "format": "text" | |
| }, | |
| { | |
| "filename": "model.bin", | |
| "format": "data" | |
| }, | |
| { | |
| "filename": "meta.json", | |
| "format": "json" | |
| } | |
| ], | |
| "corpus": [ | |
| { | |
| "NER": false, | |
| "case preserved": true, | |
| "description": "Polish CommonCrawl Dump of December 2019", | |
| "id": 117, | |
| "language": "pol", | |
| "lemmatized": false, | |
| "public": true, | |
| "stop words removal": null, | |
| "tagger": null, | |
| "tagset": null, | |
| "tokens": 32565035188, | |
| "tool": "http://dx.doi.org/10.15439/2017F432", | |
| "url": "https://commoncrawl.org/" | |
| } | |
| ], | |
| "creators": [ | |
| { | |
| "email": "kwolk@pja.edu.pl", | |
| "name": "Krzysztof Wolk" | |
| } | |
| ], | |
| "dimensions": 100, | |
| "handle": "http://vectors.nlpl.eu/repository/20/207.zip", | |
| "id": 207, | |
| "iterations": 5, | |
| "vocabulary size": 35193029, | |
| "window": 5 | |
| } |