{
  "id": 123771,
  "title": "Can't my kernel only output my submission.csv to LB?",
  "url": "/competitions/deepfake-detection-challenge/discussion/123771",
  "author_name": "",
  "post_date": "2019-12-30T09:23:29.977434500Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Hi All, \n  I've been working on my own PC to build the model and create a sumission.csv. Can't I just load the csv file to a kernel and output it to submit? I tried twice today, both poped up error and wasted my 2 submission chances....</p>\n\n<p>I really don't like the kaggle kernel function...and have been struggling with it. Any experienced guidance?</p>\n\n<p>Thanks,\n  Zanlang</p>",
  "messages": [
    {
      "id": "706351",
      "postDate": "12/30/2019 09:23:29",
      "content": "<p>Hi All, \n  I've been working on my own PC to build the model and create a sumission.csv. Can't I just load the csv file to a kernel and output it to submit? I tried twice today, both poped up error and wasted my 2 submission chances....</p>\n\n<p>I really don't like the kaggle kernel function...and have been struggling with it. Any experienced guidance?</p>\n\n<p>Thanks,\n  Zanlang</p>",
      "rawMarkdown": "Hi All, \n  I've been working on my own PC to build the model and create a sumission.csv. Can't I just load the csv file to a kernel and output it to submit? I tried twice today, both poped up error and wasted my 2 submission chances....\n\n  I really don't like the kaggle kernel function...and have been struggling with it. Any experienced guidance?\n\n  Thanks,\n  Zanlang",
      "votes": null
    },
    {
      "id": "706366",
      "postDate": "12/30/2019 09:44:28",
      "content": "<p>This won't work for this specific competition because the test set is swapped on the backend for the public leaderboard. A new set of videos and a new samplesubmission.csv file replaces the original so your old csv will not be the correct shape and will not be predictions for the correct videos. </p>",
      "rawMarkdown": "This won't work for this specific competition because the test set is swapped on the backend for the public leaderboard. A new set of videos and a new samplesubmission.csv file replaces the original so your old csv will not be the correct shape and will not be predictions for the correct videos.",
      "votes": null
    },
    {
      "id": "706821",
      "postDate": "12/30/2019 21:49:47",
      "content": "<p>Hi <a href=\"/ryches\">@ryches</a>,\nGreat thanks for your explanation. </p>\n\n<p>If my understanding is correct, what kaggle wants us to do is to use the kernel to finish the \"prediction\" step and generate a new submission csv file. This means I can develop/train my lcoal model and upload the model file to kaggle kernel to finish the prediction step there. </p>\n\n<p>Is my understanding correct?\nThanks!</p>",
      "rawMarkdown": "Hi @ryches,\nGreat thanks for your explanation. \n\nIf my understanding is correct, what kaggle wants us to do is to use the kernel to finish the \"prediction\" step and generate a new submission csv file. This means I can develop/train my lcoal model and upload the model file to kaggle kernel to finish the prediction step there. \n\nIs my understanding correct?\nThanks!",
      "votes": null
    },
    {
      "id": "706884",
      "postDate": "12/31/2019 00:02:06",
      "content": "<p>Yes, that is all correct. We are trying to do the predictions inside of their environment and there isn't any way to do a quick test with just a submission.csv unfortunately for this competition</p>",
      "rawMarkdown": "Yes, that is all correct. We are trying to do the predictions inside of their environment and there isn't any way to do a quick test with just a submission.csv unfortunately for this competition",
      "votes": null
    },
    {
      "id": "706986",
      "postDate": "12/31/2019 05:46:31",
      "content": "<p>Hi <a href=\"/ryches\">@ryches</a> rychese,\n  Great thanks for your help!</p>",
      "rawMarkdown": "Hi @ryches rychese,\n  Great thanks for your help!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 706366,
      "author_name": "ryches",
      "author_url": "",
      "post_date": "12/30/2019 09:44:28",
      "content": "<p>This won't work for this specific competition because the test set is swapped on the backend for the public leaderboard. A new set of videos and a new samplesubmission.csv file replaces the original so your old csv will not be the correct shape and will not be predictions for the correct videos. </p>",
      "votes": null,
      "replies": [
        {
          "id": 706821,
          "author_name": "",
          "author_url": "",
          "post_date": "12/30/2019 21:49:47",
          "content": "<p>Hi <a href=\"/ryches\">@ryches</a>,\nGreat thanks for your explanation. </p>\n\n<p>If my understanding is correct, what kaggle wants us to do is to use the kernel to finish the \"prediction\" step and generate a new submission csv file. This means I can develop/train my lcoal model and upload the model file to kaggle kernel to finish the prediction step there. </p>\n\n<p>Is my understanding correct?\nThanks!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 706884,
          "author_name": "ryches",
          "author_url": "",
          "post_date": "12/31/2019 00:02:06",
          "content": "<p>Yes, that is all correct. We are trying to do the predictions inside of their environment and there isn't any way to do a quick test with just a submission.csv unfortunately for this competition</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 706986,
          "author_name": "",
          "author_url": "",
          "post_date": "12/31/2019 05:46:31",
          "content": "<p>Hi <a href=\"/ryches\">@ryches</a> rychese,\n  Great thanks for your help!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "706351": "Hi All, \n  I've been working on my own PC to build the model and create a sumission.csv. Can't I just load the csv file to a kernel and output it to submit? I tried twice today, both poped up error and wasted my 2 submission chances....\n\n  I really don't like the kaggle kernel function...and have been struggling with it. Any experienced guidance?\n\n  Thanks,\n  Zanlang",
    "706366": "This won't work for this specific competition because the test set is swapped on the backend for the public leaderboard. A new set of videos and a new samplesubmission.csv file replaces the original so your old csv will not be the correct shape and will not be predictions for the correct videos.",
    "706821": "Hi @ryches,\nGreat thanks for your explanation. \n\nIf my understanding is correct, what kaggle wants us to do is to use the kernel to finish the \"prediction\" step and generate a new submission csv file. This means I can develop/train my lcoal model and upload the model file to kaggle kernel to finish the prediction step there. \n\nIs my understanding correct?\nThanks!",
    "706884": "Yes, that is all correct. We are trying to do the predictions inside of their environment and there isn't any way to do a quick test with just a submission.csv unfortunately for this competition",
    "706986": "Hi @ryches rychese,\n  Great thanks for your help!"
  },
  "source": "meta"
}