{
  "id": 181489,
  "title": "Submission score 0.0000",
  "url": "/competitions/landmark-recognition-2020/discussion/181489",
  "author_name": "",
  "post_date": "2020-09-09T05:06:32.192433700Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi, Is it mandatory to have training loops(i.e to use private train set) in submission code ? I have pre-trained model which is trained with public training set and in submission code i have only inference part for private test set, i always get 0.0000 as public score?  Can someone please help me</p>\n<p>P.S : Format of the submission file is as mentioned in evaluation tab</p>",
  "messages": [
    {
      "id": "1003564",
      "postDate": "09/09/2020 05:06:32",
      "content": "<p>Hi, Is it mandatory to have training loops(i.e to use private train set) in submission code ? I have pre-trained model which is trained with public training set and in submission code i have only inference part for private test set, i always get 0.0000 as public score?  Can someone please help me</p>\n<p>P.S : Format of the submission file is as mentioned in evaluation tab</p>",
      "rawMarkdown": "Hi, Is it mandatory to have training loops(i.e to use private train set) in submission code ? I have pre-trained model which is trained with public training set and in submission code i have only inference part for private test set, i always get 0.0000 as public score?  Can someone please help me\n\nP.S : Format of the submission file is as mentioned in evaluation tab",
      "votes": null
    },
    {
      "id": "1003610",
      "postDate": "09/09/2020 06:01:12",
      "content": "<p>No, it is not mandatory to have training loops. Actually I think it is not possible to train with the time limits for submission. If you are getting 0 as public score you might be doing inference wrong. Maybe you are just pedicting the same value for all rows in private test data. I can't tell much without seeing the code.  </p>",
      "rawMarkdown": "No, it is not mandatory to have training loops. Actually I think it is not possible to train with the time limits for submission. If you are getting 0 as public score you might be doing inference wrong. Maybe you are just pedicting the same value for all rows in private test data. I can't tell much without seeing the code.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1003610,
      "author_name": "josealways123",
      "author_url": "",
      "post_date": "09/09/2020 06:01:12",
      "content": "<p>No, it is not mandatory to have training loops. Actually I think it is not possible to train with the time limits for submission. If you are getting 0 as public score you might be doing inference wrong. Maybe you are just pedicting the same value for all rows in private test data. I can't tell much without seeing the code.  </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1003564": "Hi, Is it mandatory to have training loops(i.e to use private train set) in submission code ? I have pre-trained model which is trained with public training set and in submission code i have only inference part for private test set, i always get 0.0000 as public score?  Can someone please help me\n\nP.S : Format of the submission file is as mentioned in evaluation tab",
    "1003610": "No, it is not mandatory to have training loops. Actually I think it is not possible to train with the time limits for submission. If you are getting 0 as public score you might be doing inference wrong. Maybe you are just pedicting the same value for all rows in private test data. I can't tell much without seeing the code."
  },
  "source": "meta"
}