{
  "id": 163112,
  "title": "Is the Pub LB based only on the 10 test images?",
  "url": "/competitions/global-wheat-detection/discussion/163112",
  "author_name": "",
  "post_date": "2020-07-01T04:34:28.666955500Z",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>In the data section, I can only see 10 test images, so is the public leader-board (68% of test data) only these 10 images ?</p>",
  "messages": [
    {
      "id": "910260",
      "postDate": "07/01/2020 04:34:28",
      "content": "<p>In the data section, I can only see 10 test images, so is the public leader-board (68% of test data) only these 10 images ?</p>",
      "rawMarkdown": "In the data section, I can only see 10 test images, so is the public leader-board (68% of test data) only these 10 images ?",
      "votes": null
    },
    {
      "id": "910306",
      "postDate": "07/01/2020 05:14:13",
      "content": "<p>No, it is calculated from an unseen test set. When you hit the submit button, your kernel is re-run with the test directory replaced with different images.</p>",
      "rawMarkdown": "No, it is calculated from an unseen test set. When you hit the submit button, your kernel is re-run with the test directory replaced with different images.",
      "votes": null
    },
    {
      "id": "910390",
      "postDate": "07/01/2020 06:08:37",
      "content": "<p>Wait, really? 😮  didn't know that.</p>",
      "rawMarkdown": "Wait, really? 😮  didn't know that.",
      "votes": null
    },
    {
      "id": "910762",
      "postDate": "07/01/2020 10:58:18",
      "content": "<p>I know this happens for the private lb, so it's the same for the public LB as well. thanks</p>",
      "rawMarkdown": "I know this happens for the private lb, so it's the same for the public LB as well. thanks",
      "votes": null
    },
    {
      "id": "910845",
      "postDate": "07/01/2020 11:57:02",
      "content": "<p>look  <code>You will use more than 3,000 images from Europe (France, UK, Switzerland) and North America (Canada). The test data includes about 1,000 images from Australia, Japan, and China.</code><a href=\"https://www.kaggle.com/c/global-wheat-detection/overview\">https://www.kaggle.com/c/global-wheat-detection/overview</a>\nsome images are unseen\n<a href=\"https://www.kaggle.com/c/global-wheat-detection/overview\"></a></p>",
      "rawMarkdown": "look  `You will use more than 3,000 images from Europe (France, UK, Switzerland) and North America (Canada). The test data includes about 1,000 images from Australia, Japan, and China.`https://www.kaggle.com/c/global-wheat-detection/overview\nsome images are unseen\n[](https://www.kaggle.com/c/global-wheat-detection/overview)",
      "votes": null
    },
    {
      "id": "926264",
      "postDate": "07/12/2020 15:51:26",
      "content": "<p><a href=\"/alexanderliao\">@alexanderliao</a> If I pseudolabel and train it..and save the weights ..then did it train on all unseen data??</p>",
      "rawMarkdown": "alexanderliao If I pseudolabel and train it..and save the weights ..then did it train on all unseen data??",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 910306,
      "author_name": "alexanderliao",
      "author_url": "",
      "post_date": "07/01/2020 05:14:13",
      "content": "<p>No, it is calculated from an unseen test set. When you hit the submit button, your kernel is re-run with the test directory replaced with different images.</p>",
      "votes": null,
      "replies": [
        {
          "id": 910390,
          "author_name": "jonykarki",
          "author_url": "",
          "post_date": "07/01/2020 06:08:37",
          "content": "<p>Wait, really? 😮  didn't know that.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 910762,
          "author_name": "kennethrithvik",
          "author_url": "",
          "post_date": "07/01/2020 10:58:18",
          "content": "<p>I know this happens for the private lb, so it's the same for the public LB as well. thanks</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 926264,
          "author_name": "gyanendradas",
          "author_url": "",
          "post_date": "07/12/2020 15:51:26",
          "content": "<p><a href=\"/alexanderliao\">@alexanderliao</a> If I pseudolabel and train it..and save the weights ..then did it train on all unseen data??</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 910845,
      "author_name": "bixiaopeng",
      "author_url": "",
      "post_date": "07/01/2020 11:57:02",
      "content": "<p>look  <code>You will use more than 3,000 images from Europe (France, UK, Switzerland) and North America (Canada). The test data includes about 1,000 images from Australia, Japan, and China.</code><a href=\"https://www.kaggle.com/c/global-wheat-detection/overview\">https://www.kaggle.com/c/global-wheat-detection/overview</a>\nsome images are unseen\n<a href=\"https://www.kaggle.com/c/global-wheat-detection/overview\"></a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "910260": "In the data section, I can only see 10 test images, so is the public leader-board (68% of test data) only these 10 images ?",
    "910306": "No, it is calculated from an unseen test set. When you hit the submit button, your kernel is re-run with the test directory replaced with different images.",
    "910390": "Wait, really? 😮  didn't know that.",
    "910762": "I know this happens for the private lb, so it's the same for the public LB as well. thanks",
    "910845": "look  `You will use more than 3,000 images from Europe (France, UK, Switzerland) and North America (Canada). The test data includes about 1,000 images from Australia, Japan, and China.`https://www.kaggle.com/c/global-wheat-detection/overview\nsome images are unseen\n[](https://www.kaggle.com/c/global-wheat-detection/overview)",
    "926264": "alexanderliao If I pseudolabel and train it..and save the weights ..then did it train on all unseen data??"
  },
  "source": "meta"
}