{
  "id": 107377,
  "title": "How do you use KFolds with fastai? ",
  "url": "/competitions/aptos2019-blindness-detection/discussion/107377",
  "author_name": "",
  "post_date": "2019-09-04T03:19:41.069589800Z",
  "votes": null,
  "comment_count": 7,
  "views": 0,
  "content": "<p>I went through a lot of kernels and the documentation but I still dont seem to get it.  </p>",
  "messages": [
    {
      "id": "617342",
      "postDate": "09/04/2019 03:19:41",
      "content": "<p>I went through a lot of kernels and the documentation but I still dont seem to get it.  </p>",
      "rawMarkdown": "I went through a lot of kernels and the documentation but I still dont seem to get it.",
      "votes": null
    },
    {
      "id": "617357",
      "postDate": "09/04/2019 03:39:02",
      "content": "<p>You can try split_by_index, and use the index generated by any KFold generator.</p>",
      "rawMarkdown": "You can try split\\_by\\_index, and use the index generated by any KFold generator.",
      "votes": null
    },
    {
      "id": "618163",
      "postDate": "09/05/2019 00:24:04",
      "content": "<p>I'm interested to know this as well, afaik splitting the validation set manually into how many folds you want and then checking the score on them. </p>",
      "rawMarkdown": "I'm interested to know this as well, afaik splitting the validation set manually into how many folds you want and then checking the score on them.",
      "votes": null
    },
    {
      "id": "619203",
      "postDate": "09/06/2019 00:34:12",
      "content": "<p>I tried that and I get this error:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2Fe840194b6eb6d22110817f6e7778279a%2F1.PNG?generation=1567730020911034&amp;alt=media\" alt=\"\">\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2F2830083360e3b14f0766b133d32c5577%2F3.PNG?generation=1567730132673685&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "I tried that and I get this error:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2Fe840194b6eb6d22110817f6e7778279a%2F1.PNG?generation=1567730020911034&amp;alt=media)\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2F2830083360e3b14f0766b133d32c5577%2F3.PNG?generation=1567730132673685&amp;alt=media)",
      "votes": null
    },
    {
      "id": "619205",
      "postDate": "09/06/2019 00:35:00",
      "content": "<p>Not sure really on how to fix it.</p>",
      "rawMarkdown": "Not sure really on how to fix it.",
      "votes": null
    },
    {
      "id": "619329",
      "postDate": "09/06/2019 04:51:13",
      "content": "<p>You need to pass in the train index and test index in split by index! You can make a loop and train on each fold one by one</p>",
      "rawMarkdown": "You need to pass in the train index and test index in split by index! You can make a loop and train on each fold one by one",
      "votes": null
    },
    {
      "id": "620040",
      "postDate": "09/06/2019 22:46:11",
      "content": "<p>I am not sure what that means. Could you help me or explain it? It would be greatly appreciated. Thank you!</p>",
      "rawMarkdown": "I am not sure what that means. Could you help me or explain it? It would be greatly appreciated. Thank you!",
      "votes": null
    },
    {
      "id": "620229",
      "postDate": "09/07/2019 07:15:31",
      "content": "<p>You could refer to this awesome kernal by ilovescience\n<a href=\"https://www.kaggle.com/tanlikesmath/aptos-2019-previous-dr-efficientnetb3-single-model\">https://www.kaggle.com/tanlikesmath/aptos-2019-previous-dr-efficientnetb3-single-model</a></p>",
      "rawMarkdown": "You could refer to this awesome kernal by ilovescience\nhttps://www.kaggle.com/tanlikesmath/aptos-2019-previous-dr-efficientnetb3-single-model",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 617357,
      "author_name": "kenho211",
      "author_url": "",
      "post_date": "09/04/2019 03:39:02",
      "content": "<p>You can try split_by_index, and use the index generated by any KFold generator.</p>",
      "votes": null,
      "replies": [
        {
          "id": 619203,
          "author_name": "krsoju",
          "author_url": "",
          "post_date": "09/06/2019 00:34:12",
          "content": "<p>I tried that and I get this error:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2Fe840194b6eb6d22110817f6e7778279a%2F1.PNG?generation=1567730020911034&amp;alt=media\" alt=\"\">\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2F2830083360e3b14f0766b133d32c5577%2F3.PNG?generation=1567730132673685&amp;alt=media\" alt=\"\"></p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 619205,
          "author_name": "krsoju",
          "author_url": "",
          "post_date": "09/06/2019 00:35:00",
          "content": "<p>Not sure really on how to fix it.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 619329,
          "author_name": "ubamba98",
          "author_url": "",
          "post_date": "09/06/2019 04:51:13",
          "content": "<p>You need to pass in the train index and test index in split by index! You can make a loop and train on each fold one by one</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 620040,
          "author_name": "krsoju",
          "author_url": "",
          "post_date": "09/06/2019 22:46:11",
          "content": "<p>I am not sure what that means. Could you help me or explain it? It would be greatly appreciated. Thank you!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 620229,
          "author_name": "ubamba98",
          "author_url": "",
          "post_date": "09/07/2019 07:15:31",
          "content": "<p>You could refer to this awesome kernal by ilovescience\n<a href=\"https://www.kaggle.com/tanlikesmath/aptos-2019-previous-dr-efficientnetb3-single-model\">https://www.kaggle.com/tanlikesmath/aptos-2019-previous-dr-efficientnetb3-single-model</a></p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 618163,
      "author_name": "r20367585",
      "author_url": "",
      "post_date": "09/05/2019 00:24:04",
      "content": "<p>I'm interested to know this as well, afaik splitting the validation set manually into how many folds you want and then checking the score on them. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "617342": "I went through a lot of kernels and the documentation but I still dont seem to get it.",
    "617357": "You can try split\\_by\\_index, and use the index generated by any KFold generator.",
    "618163": "I'm interested to know this as well, afaik splitting the validation set manually into how many folds you want and then checking the score on them.",
    "619203": "I tried that and I get this error:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2Fe840194b6eb6d22110817f6e7778279a%2F1.PNG?generation=1567730020911034&amp;alt=media)\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2448809%2F2830083360e3b14f0766b133d32c5577%2F3.PNG?generation=1567730132673685&amp;alt=media)",
    "619205": "Not sure really on how to fix it.",
    "619329": "You need to pass in the train index and test index in split by index! You can make a loop and train on each fold one by one",
    "620040": "I am not sure what that means. Could you help me or explain it? It would be greatly appreciated. Thank you!",
    "620229": "You could refer to this awesome kernal by ilovescience\nhttps://www.kaggle.com/tanlikesmath/aptos-2019-previous-dr-efficientnetb3-single-model"
  },
  "source": "meta"
}