{
  "id": 142330,
  "title": "Yet another multi-lingual model",
  "url": "/competitions/jigsaw-multilingual-toxic-comment-classification/discussion/142330",
  "author_name": "ilovescience",
  "post_date": "2020-04-10T01:36:51.799000",
  "votes": 4,
  "comment_count": 0,
  "views": 0,
  "content": "<p>This model developed by Microsoft is distilled from the XLM-R model, so it's worse than XLM-R but better than mBERT:</p>\n\n<p><a href=\"https://arxiv.org/pdf/2002.10957.pdf\">https://arxiv.org/pdf/2002.10957.pdf</a></p>\n\n<p>Could be something useful to add to our ensembles?</p>",
  "messages": [
    {
      "id": 802950,
      "postDate": "2020-04-10T01:36:51.800Z",
      "content": "<p>This model developed by Microsoft is distilled from the XLM-R model, so it's worse than XLM-R but better than mBERT:</p>\n\n<p><a href=\"https://arxiv.org/pdf/2002.10957.pdf\">https://arxiv.org/pdf/2002.10957.pdf</a></p>\n\n<p>Could be something useful to add to our ensembles?</p>",
      "rawMarkdown": "This model developed by Microsoft is distilled from the XLM-R model, so it's worse than XLM-R but better than mBERT:\n\nhttps://arxiv.org/pdf/2002.10957.pdf\n\nCould be something useful to add to our ensembles?",
      "votes": 4
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "802950": "This model developed by Microsoft is distilled from the XLM-R model, so it's worse than XLM-R but better than mBERT:\n\nhttps://arxiv.org/pdf/2002.10957.pdf\n\nCould be something useful to add to our ensembles?"
  }
}