{
  "id": 162572,
  "title": "What is the size of the final evaluation dataset?",
  "url": "/competitions/global-wheat-detection/discussion/162572",
  "author_name": "",
  "post_date": "2020-06-29T12:13:24.198483900Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Dear fellow competitors and hosts,\nI believe it would be beneficial for all parties involved if the hosts could let us know the <strong>number of images the final evaluation</strong> of our models will be done on. \nThe current restrictions for code submissions dictate that &lt;6h of GPU run-time is allowed to be used. This restriction, along with the number of images for final evaluation will be a strong guideline for us to decide how heavy/extensive we want our model ensembling to be. </p>\n\n<p>Currently, not knowing this, leaves us in a position where we could undertake heavy ensembling which will provide significantly better results, but might fail to meet the final 6h runtime requirement.</p>",
  "messages": [
    {
      "id": "906556",
      "postDate": "06/29/2020 12:13:24",
      "content": "<p>Dear fellow competitors and hosts,\nI believe it would be beneficial for all parties involved if the hosts could let us know the <strong>number of images the final evaluation</strong> of our models will be done on. \nThe current restrictions for code submissions dictate that &lt;6h of GPU run-time is allowed to be used. This restriction, along with the number of images for final evaluation will be a strong guideline for us to decide how heavy/extensive we want our model ensembling to be. </p>\n\n<p>Currently, not knowing this, leaves us in a position where we could undertake heavy ensembling which will provide significantly better results, but might fail to meet the final 6h runtime requirement.</p>",
      "rawMarkdown": "Dear fellow competitors and hosts,\nI believe it would be beneficial for all parties involved if the hosts could let us know the **number of images the final evaluation** of our models will be done on. \nThe current restrictions for code submissions dictate that &lt;6h of GPU run-time is allowed to be used. This restriction, along with the number of images for final evaluation will be a strong guideline for us to decide how heavy/extensive we want our model ensembling to be. \n\nCurrently, not knowing this, leaves us in a position where we could undertake heavy ensembling which will provide significantly better results, but might fail to meet the final 6h runtime requirement.",
      "votes": null
    },
    {
      "id": "906602",
      "postDate": "06/29/2020 12:45:41",
      "content": "<p>It's all explained <a href=\"https://www.kaggle.com/c/global-wheat-detection/discussion/161081\">here</a>.</p>",
      "rawMarkdown": "It's all explained [here](https://www.kaggle.com/c/global-wheat-detection/discussion/161081).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 906602,
      "author_name": "matthewmasters",
      "author_url": "",
      "post_date": "06/29/2020 12:45:41",
      "content": "<p>It's all explained <a href=\"https://www.kaggle.com/c/global-wheat-detection/discussion/161081\">here</a>.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "906556": "Dear fellow competitors and hosts,\nI believe it would be beneficial for all parties involved if the hosts could let us know the **number of images the final evaluation** of our models will be done on. \nThe current restrictions for code submissions dictate that &lt;6h of GPU run-time is allowed to be used. This restriction, along with the number of images for final evaluation will be a strong guideline for us to decide how heavy/extensive we want our model ensembling to be. \n\nCurrently, not knowing this, leaves us in a position where we could undertake heavy ensembling which will provide significantly better results, but might fail to meet the final 6h runtime requirement.",
    "906602": "It's all explained [here](https://www.kaggle.com/c/global-wheat-detection/discussion/161081)."
  },
  "source": "meta"
}