{
  "id": 185294,
  "title": "Really dumb question: how to attach the full train set?",
  "url": "/competitions/landmark-recognition-2020/discussion/185294",
  "author_name": "",
  "post_date": "2020-09-20T09:13:06.306670400Z",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>The data description says:</p>\n<blockquote>\n  <p>To facilitate recognition-by-retrieval approaches, the private training set contains only a 100k subset of the total public training set. This 100k subset contains all of the training set images associated with the landmarks in the private test set. <strong>You may still attach the full training set as an external data set if you wish.</strong></p>\n</blockquote>\n<p>Sooo… how to attach the full training set? As I understand, the <code>../input/landmark-recognition-2020</code> directory will be somehow replaced with the private train images during re-run. </p>",
  "messages": [
    {
      "id": "1019197",
      "postDate": "09/20/2020 09:13:06",
      "content": "<p>The data description says:</p>\n<blockquote>\n  <p>To facilitate recognition-by-retrieval approaches, the private training set contains only a 100k subset of the total public training set. This 100k subset contains all of the training set images associated with the landmarks in the private test set. <strong>You may still attach the full training set as an external data set if you wish.</strong></p>\n</blockquote>\n<p>Sooo… how to attach the full training set? As I understand, the <code>../input/landmark-recognition-2020</code> directory will be somehow replaced with the private train images during re-run. </p>",
      "rawMarkdown": "The data description says:\n> To facilitate recognition-by-retrieval approaches, the private training set contains only a 100k subset of the total public training set. This 100k subset contains all of the training set images associated with the landmarks in the private test set. **You may still attach the full training set as an external data set if you wish.**\n\nSooo... how to attach the full training set? As I understand, the `../input/landmark-recognition-2020` directory will be somehow replaced with the private train images during re-run.",
      "votes": null
    },
    {
      "id": "1019428",
      "postDate": "09/20/2020 12:20:18",
      "content": "<p>upload the train.csv as additional dataset and link in kernel</p>",
      "rawMarkdown": "upload the train.csv as additional dataset and link in kernel",
      "votes": null
    },
    {
      "id": "1019528",
      "postDate": "09/20/2020 14:00:29",
      "content": "<p>Oh, that's easier than I expected! Thanks :D</p>",
      "rawMarkdown": "Oh, that's easier than I expected! Thanks :D",
      "votes": null
    },
    {
      "id": "1019595",
      "postDate": "09/20/2020 14:30:24",
      "content": "<p>I mean you would need to upload the full competiton data as a dataset not just the csv if you want to use the \"public\" train images. </p>",
      "rawMarkdown": "I mean you would need to upload the full competiton data as a dataset not just the csv if you want to use the \"public\" train images.",
      "votes": null
    },
    {
      "id": "1020052",
      "postDate": "09/20/2020 21:08:52",
      "content": "<p><a href=\"https://www.kaggle.com/christofhenkel\" target=\"_blank\">@christofhenkel</a> sooooo… the only way is to split the dataset into chunks of 20GB (so 5 datasets), upload them, and attach to the notebook?</p>",
      "rawMarkdown": "christofhenkel sooooo... the only way is to split the dataset into chunks of 20GB (so 5 datasets), upload them, and attach to the notebook?",
      "votes": null
    },
    {
      "id": "1020268",
      "postDate": "09/21/2020 04:14:52",
      "content": "<p>probably   </p>",
      "rawMarkdown": "probably",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1019428,
      "author_name": "christofhenkel",
      "author_url": "",
      "post_date": "09/20/2020 12:20:18",
      "content": "<p>upload the train.csv as additional dataset and link in kernel</p>",
      "votes": null,
      "replies": [
        {
          "id": 1019528,
          "author_name": "chankhavu",
          "author_url": "",
          "post_date": "09/20/2020 14:00:29",
          "content": "<p>Oh, that's easier than I expected! Thanks :D</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1019595,
          "author_name": "christofhenkel",
          "author_url": "",
          "post_date": "09/20/2020 14:30:24",
          "content": "<p>I mean you would need to upload the full competiton data as a dataset not just the csv if you want to use the \"public\" train images. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1020052,
          "author_name": "chankhavu",
          "author_url": "",
          "post_date": "09/20/2020 21:08:52",
          "content": "<p><a href=\"https://www.kaggle.com/christofhenkel\" target=\"_blank\">@christofhenkel</a> sooooo… the only way is to split the dataset into chunks of 20GB (so 5 datasets), upload them, and attach to the notebook?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1020268,
          "author_name": "christofhenkel",
          "author_url": "",
          "post_date": "09/21/2020 04:14:52",
          "content": "<p>probably   </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1019197": "The data description says:\n> To facilitate recognition-by-retrieval approaches, the private training set contains only a 100k subset of the total public training set. This 100k subset contains all of the training set images associated with the landmarks in the private test set. **You may still attach the full training set as an external data set if you wish.**\n\nSooo... how to attach the full training set? As I understand, the `../input/landmark-recognition-2020` directory will be somehow replaced with the private train images during re-run.",
    "1019428": "upload the train.csv as additional dataset and link in kernel",
    "1019528": "Oh, that's easier than I expected! Thanks :D",
    "1019595": "I mean you would need to upload the full competiton data as a dataset not just the csv if you want to use the \"public\" train images.",
    "1020052": "christofhenkel sooooo... the only way is to split the dataset into chunks of 20GB (so 5 datasets), upload them, and attach to the notebook?",
    "1020268": "probably"
  },
  "source": "meta"
}