{
  "id": 30737,
  "title": "Validation and leaderboard score inconsistencies",
  "url": "/competitions/intel-mobileodt-cervical-cancer-screening/discussion/30737",
  "author_name": "",
  "post_date": "2017-03-27T18:25:07.673225400Z",
  "votes": null,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I'm seeing a steadily climbing validation score as I refine my models and data augmentation, but a steadily declining public leaderboard score. Anyone else seeing this?</p>",
  "messages": [
    {
      "id": "170824",
      "postDate": "03/27/2017 18:25:07",
      "content": "<p>I'm seeing a steadily climbing validation score as I refine my models and data augmentation, but a steadily declining public leaderboard score. Anyone else seeing this?</p>",
      "rawMarkdown": "I'm seeing a steadily climbing validation score as I refine my models and data augmentation, but a steadily declining public leaderboard score. Anyone else seeing this?",
      "votes": null
    },
    {
      "id": "170992",
      "postDate": "03/28/2017 09:03:15",
      "content": "<p>How are you performing validation?</p>",
      "rawMarkdown": "How are you performing validation?",
      "votes": null
    },
    {
      "id": "172999",
      "postDate": "04/05/2017 15:15:36",
      "content": "<p>Yup. I am seeing the same phenomenon and looking at the performance graphs for loss and accuracy. There is a steady \"learning\" that seems to be happening with the augmented training data. However, the better my models perform on the training data, the worse it gets on the test data. </p>",
      "rawMarkdown": "Yup. I am seeing the same phenomenon and looking at the performance graphs for loss and accuracy. There is a steady \"learning\" that seems to be happening with the augmented training data. However, the better my models perform on the training data, the worse it gets on the test data.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 170992,
      "author_name": "yadavsarthak",
      "author_url": "",
      "post_date": "03/28/2017 09:03:15",
      "content": "<p>How are you performing validation?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 172999,
      "author_name": "tagzydeco",
      "author_url": "",
      "post_date": "04/05/2017 15:15:36",
      "content": "<p>Yup. I am seeing the same phenomenon and looking at the performance graphs for loss and accuracy. There is a steady \"learning\" that seems to be happening with the augmented training data. However, the better my models perform on the training data, the worse it gets on the test data. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "170824": "I'm seeing a steadily climbing validation score as I refine my models and data augmentation, but a steadily declining public leaderboard score. Anyone else seeing this?",
    "170992": "How are you performing validation?",
    "172999": "Yup. I am seeing the same phenomenon and looking at the performance graphs for loss and accuracy. There is a steady \"learning\" that seems to be happening with the augmented training data. However, the better my models perform on the training data, the worse it gets on the test data."
  },
  "source": "meta"
}