{
  "id": 269801,
  "title": "different results in local and kaggle",
  "url": "/competitions/rsna-miccai-brain-tumor-radiogenomic-classification/discussion/269801",
  "author_name": "",
  "post_date": "2021-09-02T06:30:46.316622400Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>i am unable to reproduce results of local environment on kaggle, trained and tested on local and finally testing on kaggle(not training) but the result are not same, any clue whats going on, can i reproduce my env on kaggle via docker? or something. </p>",
  "messages": [
    {
      "id": "1500067",
      "postDate": "09/02/2021 06:30:46",
      "content": "<p>i am unable to reproduce results of local environment on kaggle, trained and tested on local and finally testing on kaggle(not training) but the result are not same, any clue whats going on, can i reproduce my env on kaggle via docker? or something. </p>",
      "rawMarkdown": "i am unable to reproduce results of local environment on kaggle, trained and tested on local and finally testing on kaggle(not training) but the result are not same, any clue whats going on, can i reproduce my env on kaggle via docker? or something.",
      "votes": null
    },
    {
      "id": "1500469",
      "postDate": "09/02/2021 12:07:58",
      "content": "<p>Are you testing against data with known labels or against test data?</p>",
      "rawMarkdown": "Are you testing against data with known labels or against test data?",
      "votes": null
    },
    {
      "id": "1501254",
      "postDate": "09/03/2021 04:22:35",
      "content": "<p>known labels, a split of train data</p>",
      "rawMarkdown": "known labels, a split of train data",
      "votes": null
    },
    {
      "id": "1504044",
      "postDate": "09/06/2021 03:05:53",
      "content": "<p>Not sure if you're doing this but if you're comparing with the leaderboard results then auc is very volatile for small datasets. </p>",
      "rawMarkdown": "Not sure if you're doing this but if you're comparing with the leaderboard results then auc is very volatile for small datasets.",
      "votes": null
    },
    {
      "id": "1504278",
      "postDate": "09/06/2021 08:29:46",
      "content": "<p>GPU difference may be one reason.</p>",
      "rawMarkdown": "GPU difference may be one reason.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1500469,
      "author_name": "aryamansharma47",
      "author_url": "",
      "post_date": "09/02/2021 12:07:58",
      "content": "<p>Are you testing against data with known labels or against test data?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1501254,
      "author_name": "shummi39",
      "author_url": "",
      "post_date": "09/03/2021 04:22:35",
      "content": "<p>known labels, a split of train data</p>",
      "votes": null,
      "replies": [
        {
          "id": 1504044,
          "author_name": "aryamansharma47",
          "author_url": "",
          "post_date": "09/06/2021 03:05:53",
          "content": "<p>Not sure if you're doing this but if you're comparing with the leaderboard results then auc is very volatile for small datasets. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1504278,
      "author_name": "drtausamaru",
      "author_url": "",
      "post_date": "09/06/2021 08:29:46",
      "content": "<p>GPU difference may be one reason.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1500067": "i am unable to reproduce results of local environment on kaggle, trained and tested on local and finally testing on kaggle(not training) but the result are not same, any clue whats going on, can i reproduce my env on kaggle via docker? or something.",
    "1500469": "Are you testing against data with known labels or against test data?",
    "1501254": "known labels, a split of train data",
    "1504044": "Not sure if you're doing this but if you're comparing with the leaderboard results then auc is very volatile for small datasets.",
    "1504278": "GPU difference may be one reason."
  },
  "source": "meta"
}