{
  "id": 440878,
  "title": "New to Kaggle, why are we not able to get same score from the notebook we copied and working on",
  "url": "/competitions/predict-ai-model-runtime/discussion/440878",
  "author_name": "",
  "post_date": "2023-09-16T15:09:55.801054900Z",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default (I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default(as the authors used) parameters and models. I wonder why ? </p>",
  "messages": [
    {
      "id": "2441940",
      "postDate": "09/16/2023 15:09:55",
      "content": "<p>I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default (I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default(as the authors used) parameters and models. I wonder why ? </p>",
      "rawMarkdown": "I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default (I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default(as the authors used) parameters and models. I wonder why ?",
      "votes": null
    },
    {
      "id": "2441945",
      "postDate": "09/16/2023 15:16:39",
      "content": "<p>Typically, multiple versions of each uploaded notebook are available, and I believe that the scores are derived from the best version. This best version is often the latest one or the one that's pinned, although it's not always the case. Therefore, I recommend verifying the score associated with the specific version.</p>",
      "rawMarkdown": "Typically, multiple versions of each uploaded notebook are available, and I believe that the scores are derived from the best version. This best version is often the latest one or the one that's pinned, although it's not always the case. Therefore, I recommend verifying the score associated with the specific version.",
      "votes": null
    },
    {
      "id": "2441951",
      "postDate": "09/16/2023 15:26:23",
      "content": "<p>The models may not be seeded, hence different executions may lead to slightly different outcomes.</p>",
      "rawMarkdown": "The models may not be seeded, hence different executions may lead to slightly different outcomes.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2441945,
      "author_name": "annafabris",
      "author_url": "",
      "post_date": "09/16/2023 15:16:39",
      "content": "<p>Typically, multiple versions of each uploaded notebook are available, and I believe that the scores are derived from the best version. This best version is often the latest one or the one that's pinned, although it's not always the case. Therefore, I recommend verifying the score associated with the specific version.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2441951,
      "author_name": "fabiendaniel",
      "author_url": "",
      "post_date": "09/16/2023 15:26:23",
      "content": "<p>The models may not be seeded, hence different executions may lead to slightly different outcomes.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2441940": "I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default (I've tried to improve the baselines that people have built and shared on the code sections, but it seems like most of the time we won't be able to get the baseline score from the default(as the authors used) parameters and models. I wonder why ?",
    "2441945": "Typically, multiple versions of each uploaded notebook are available, and I believe that the scores are derived from the best version. This best version is often the latest one or the one that's pinned, although it's not always the case. Therefore, I recommend verifying the score associated with the specific version.",
    "2441951": "The models may not be seeded, hence different executions may lead to slightly different outcomes."
  },
  "source": "meta"
}