|
"description": "CCMatrix: Mining Billions of High-Quality Parallel Sentences on the WEB\n\nWe show that margin-based bitext mining in LASER's multilingual sentence space can be applied to\nmonolingual corpora of billions of sentences to produce high quality aligned translation data.\nWe use thirty-two snapshots of a curated common crawl corpus [1] totaling 69 billion unique sentences.\nUsing one unified approach for 80 languages, we were able to mine 10.8 billion parallel sentences,\nout of which only 2.9 billion are aligned with English.\n\nIMPORTANT: Please cite reference [2][3] if you use this data.\n\n[1] Guillaume Wenzek, Marie-Anne Lachaux, Alexis Conneau, Vishrav Chaudhary, Francisco Guzm\u00e1n, Armand Jouli\n and Edouard Grave, CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data\n\n[2] Holger Schwenk, Guillaume Wenzek, Sergey Edunov, Edouard Grave and Armand Joulin,\n CCMatrix: Mining Billions of High-Quality Parallel Sentences on the WEB\n\n[3] Angela Fan, Shruti Bhosale, Holger Schwenk, Zhiyi Ma, Ahmed El-Kishky, Siddharth Goyal, Mandeep Baines,\n Onur Celebi, Guillaume Wenzek, Vishrav Chaudhary, Naman Goyal, Tom Birch, Vitaliy Liptchinsky,\n Sergey Edunov, Edouard Grave, Michael Auli, and Armand Joulin.\n Beyond English-Centric Multilingual Machine Translation\n \n90 languages, 1,197 bitexts\ntotal number of files: 90\ntotal number of tokens: 112.14G\ntotal number of sentence fragments: 7.37G", |
|
"citation": "Guillaume Wenzek, Marie-Anne Lachaux, Alexis Conneau, Vishrav Chaudhary, Francisco Guzm\u00e1n, Armand Jouli and Edouard Grave, CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data", |
