{
  "id": 274716,
  "title": "Use pretrained embeddings",
  "url": "/competitions/landmark-recognition-2021/discussion/274716",
  "author_name": "",
  "post_date": "2021-09-27T13:43:44.343913600Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>So we cant generate all embeddings for train, save them and then make generate predictions?</p>",
  "messages": [
    {
      "id": "1525541",
      "postDate": "09/27/2021 13:43:44",
      "content": "<p>So we cant generate all embeddings for train, save them and then make generate predictions?</p>",
      "rawMarkdown": "So we cant generate all embeddings for train, save them and then make generate predictions?",
      "votes": null
    },
    {
      "id": "1526842",
      "postDate": "09/28/2021 11:26:15",
      "content": "<p>Yes, but the saved weights should be available for all in Kaggle/GitHub as a public dataset.</p>",
      "rawMarkdown": "Yes, but the saved weights should be available for all in Kaggle/GitHub as a public dataset.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1526842,
      "author_name": "markbquant",
      "author_url": "",
      "post_date": "09/28/2021 11:26:15",
      "content": "<p>Yes, but the saved weights should be available for all in Kaggle/GitHub as a public dataset.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1525541": "So we cant generate all embeddings for train, save them and then make generate predictions?",
    "1526842": "Yes, but the saved weights should be available for all in Kaggle/GitHub as a public dataset."
  },
  "source": "meta"
}