{
  "id": 670607,
  "title": "Final Competition Leaderboard",
  "url": "/competitions/csiro-biomass/discussion/670607",
  "author_name": "",
  "post_date": "2026-01-28T20:24:11.408156600Z",
  "votes": null,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi All, </p>\n<p>Will the final competition lb be calculated on submissions where the sub has full access to the models and training data, i.e. will it remove all of the subs that have just copy and pasted other code without access to the models? </p>\n<p>Wishing everyone all the best for the final scoring. I know I had great fun throughout this challenge.</p>",
  "messages": [
    {
      "id": "3398213",
      "postDate": "01/28/2026 20:24:11",
      "content": "<p>Hi All, </p>\n<p>Will the final competition lb be calculated on submissions where the sub has full access to the models and training data, i.e. will it remove all of the subs that have just copy and pasted other code without access to the models? </p>\n<p>Wishing everyone all the best for the final scoring. I know I had great fun throughout this challenge.</p>",
      "rawMarkdown": "Hi All, \n\nWill the final competition lb be calculated on submissions where the sub has full access to the models and training data, i.e. will it remove all of the subs that have just copy and pasted other code without access to the models? \n\nWishing everyone all the best for the final scoring. I know I had great fun throughout this challenge.",
      "votes": null
    },
    {
      "id": "3398234",
      "postDate": "01/28/2026 20:57:09",
      "content": "<p>I believe the system filters those out automatically. If a submission doesn't have access to the required files, it typically results in a 'Submission Error' and doesn't get a rank on the leaderboard and public sharing is not a problem in kaggle ig the only thing that can happen is that many public notebooks have some magic numbers post processing tricks which can drastically change the current leaderboard scenario\nHappy kaggling!! Good luck for the private lb🫠</p>",
      "rawMarkdown": "I believe the system filters those out automatically. If a submission doesn't have access to the required files, it typically results in a 'Submission Error' and doesn't get a rank on the leaderboard and public sharing is not a problem in kaggle ig the only thing that can happen is that many public notebooks have some magic numbers post processing tricks which can drastically change the current leaderboard scenario\nHappy kaggling!! Good luck for the private lb🫠",
      "votes": null
    },
    {
      "id": "3398240",
      "postDate": "01/28/2026 21:30:12",
      "content": "<p>If that is true, many submissions will fail. There is a notebook called CSIRO-Data Split that computes image embeddings. The notebook contains a private dataset which is used but its output is accessible. It is part of a public notebook that scores 0.72. So if all submissions that use this notebook output are to be eliminated, I wonder what will happen 😆</p>",
      "rawMarkdown": "If that is true, many submissions will fail. There is a notebook called CSIRO-Data Split that computes image embeddings. The notebook contains a private dataset which is used but its output is accessible. It is part of a public notebook that scores 0.72. So if all submissions that use this notebook output are to be eliminated, I wonder what will happen 😆",
      "votes": null
    },
    {
      "id": "3398242",
      "postDate": "01/28/2026 21:34:11",
      "content": "<p>Yes I was wondering about those with private datasets, it would technically work for the leaderboard submission but will it work for the final comp. It'll give the final LB a good clean if it blocks all of those.</p>",
      "rawMarkdown": "Yes I was wondering about those with private datasets, it would technically work for the leaderboard submission but will it work for the final comp. It'll give the final LB a good clean if it blocks all of those.",
      "votes": null
    },
    {
      "id": "3398245",
      "postDate": "01/28/2026 21:42:13",
      "content": "<p>If u have a private dataset that is being used in the notebook and is public u can fork it but u wont br able to generate a submission with it because that dataset doesn't appear in the environment for u </p>",
      "rawMarkdown": "If u have a private dataset that is being used in the notebook and is public u can fork it but u wont br able to generate a submission with it because that dataset doesn't appear in the environment for u",
      "votes": null
    },
    {
      "id": "3398246",
      "postDate": "01/28/2026 21:43:10",
      "content": "<p>That notebook yup the siglip one btw what's the highest non dino model score could u reach?  Or any machine learning approach?</p>",
      "rawMarkdown": "That notebook yup the siglip one btw what's the highest non dino model score could u reach?  Or any machine learning approach?",
      "votes": null
    },
    {
      "id": "3398259",
      "postDate": "01/28/2026 22:34:27",
      "content": "<p>The private LB is already computed at submission time, just like the public one Kaggle doesn’t re-run notebooks or re-check data access after the competition ends. If a submission was valid and scored, its private score is already locked in.</p>\n<p>Post-competition runtime checks usually happen only in specific cases (e.g., time-series, code competitions, or large challenges) and only when explicitly stated in the rules. Otherwise, the final LB is simply the reveal of the hidden private scores.</p>",
      "rawMarkdown": "The private LB is already computed at submission time, just like the public one Kaggle doesn’t re-run notebooks or re-check data access after the competition ends. If a submission was valid and scored, its private score is already locked in.\n\nPost-competition runtime checks usually happen only in specific cases (e.g., time-series, code competitions, or large challenges) and only when explicitly stated in the rules. Otherwise, the final LB is simply the reveal of the hidden private scores.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3398234,
      "author_name": "suman2208",
      "author_url": "",
      "post_date": "01/28/2026 20:57:09",
      "content": "<p>I believe the system filters those out automatically. If a submission doesn't have access to the required files, it typically results in a 'Submission Error' and doesn't get a rank on the leaderboard and public sharing is not a problem in kaggle ig the only thing that can happen is that many public notebooks have some magic numbers post processing tricks which can drastically change the current leaderboard scenario\nHappy kaggling!! Good luck for the private lb🫠</p>",
      "votes": null,
      "replies": [
        {
          "id": 3398240,
          "author_name": "jeannkouagou",
          "author_url": "",
          "post_date": "01/28/2026 21:30:12",
          "content": "<p>If that is true, many submissions will fail. There is a notebook called CSIRO-Data Split that computes image embeddings. The notebook contains a private dataset which is used but its output is accessible. It is part of a public notebook that scores 0.72. So if all submissions that use this notebook output are to be eliminated, I wonder what will happen 😆</p>",
          "votes": null,
          "replies": [
            {
              "id": 3398242,
              "author_name": "alloyd256",
              "author_url": "",
              "post_date": "01/28/2026 21:34:11",
              "content": "<p>Yes I was wondering about those with private datasets, it would technically work for the leaderboard submission but will it work for the final comp. It'll give the final LB a good clean if it blocks all of those.</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3398245,
                  "author_name": "suman2208",
                  "author_url": "",
                  "post_date": "01/28/2026 21:42:13",
                  "content": "<p>If u have a private dataset that is being used in the notebook and is public u can fork it but u wont br able to generate a submission with it because that dataset doesn't appear in the environment for u </p>",
                  "votes": null,
                  "replies": []
                }
              ]
            },
            {
              "id": 3398246,
              "author_name": "suman2208",
              "author_url": "",
              "post_date": "01/28/2026 21:43:10",
              "content": "<p>That notebook yup the siglip one btw what's the highest non dino model score could u reach?  Or any machine learning approach?</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 3398259,
      "author_name": "",
      "author_url": "",
      "post_date": "01/28/2026 22:34:27",
      "content": "<p>The private LB is already computed at submission time, just like the public one Kaggle doesn’t re-run notebooks or re-check data access after the competition ends. If a submission was valid and scored, its private score is already locked in.</p>\n<p>Post-competition runtime checks usually happen only in specific cases (e.g., time-series, code competitions, or large challenges) and only when explicitly stated in the rules. Otherwise, the final LB is simply the reveal of the hidden private scores.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3398213": "Hi All, \n\nWill the final competition lb be calculated on submissions where the sub has full access to the models and training data, i.e. will it remove all of the subs that have just copy and pasted other code without access to the models? \n\nWishing everyone all the best for the final scoring. I know I had great fun throughout this challenge.",
    "3398234": "I believe the system filters those out automatically. If a submission doesn't have access to the required files, it typically results in a 'Submission Error' and doesn't get a rank on the leaderboard and public sharing is not a problem in kaggle ig the only thing that can happen is that many public notebooks have some magic numbers post processing tricks which can drastically change the current leaderboard scenario\nHappy kaggling!! Good luck for the private lb🫠",
    "3398240": "If that is true, many submissions will fail. There is a notebook called CSIRO-Data Split that computes image embeddings. The notebook contains a private dataset which is used but its output is accessible. It is part of a public notebook that scores 0.72. So if all submissions that use this notebook output are to be eliminated, I wonder what will happen 😆",
    "3398242": "Yes I was wondering about those with private datasets, it would technically work for the leaderboard submission but will it work for the final comp. It'll give the final LB a good clean if it blocks all of those.",
    "3398245": "If u have a private dataset that is being used in the notebook and is public u can fork it but u wont br able to generate a submission with it because that dataset doesn't appear in the environment for u",
    "3398246": "That notebook yup the siglip one btw what's the highest non dino model score could u reach?  Or any machine learning approach?",
    "3398259": "The private LB is already computed at submission time, just like the public one Kaggle doesn’t re-run notebooks or re-check data access after the competition ends. If a submission was valid and scored, its private score is already locked in.\n\nPost-competition runtime checks usually happen only in specific cases (e.g., time-series, code competitions, or large challenges) and only when explicitly stated in the rules. Otherwise, the final LB is simply the reveal of the hidden private scores."
  },
  "source": "meta"
}