{
  "id": 613091,
  "title": "pairs of authentic ans forged images in the test set",
  "url": "/competitions/recodai-luc-scientific-image-forgery-detection/discussion/613091",
  "author_name": "",
  "post_date": "2025-10-24T05:54:11.928562200Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi, i whish to know if there are always  pairs of images (authentic and forged)  that could be shufffled in the test set?</p>",
  "messages": [
    {
      "id": "3306221",
      "postDate": "10/24/2025 05:54:11",
      "content": "<p>Hi, i whish to know if there are always  pairs of images (authentic and forged)  that could be shufffled in the test set?</p>",
      "rawMarkdown": "Hi, i whish to know if there are always  pairs of images (authentic and forged)  that could be shufffled in the test set?",
      "votes": null
    },
    {
      "id": "3306338",
      "postDate": "10/24/2025 10:45:46",
      "content": "<p>That's what your model has to predict. And mask for the ones your model determine are forged.</p>",
      "rawMarkdown": "That's what your model has to predict. And mask for the ones your model determine are forged.",
      "votes": null
    },
    {
      "id": "3307014",
      "postDate": "10/26/2025 00:28:18",
      "content": "<p>any submit code available ?</p>",
      "rawMarkdown": "any submit code available ?",
      "votes": null
    },
    {
      "id": "3308088",
      "postDate": "10/28/2025 14:36:32",
      "content": "<p>Hi, if there were such pairs, then you would get a big data leak during training and that would make your validation score unreliable. Hence if that was the case, I would recommend deleting the ones that do have a duplicate in the forged dataset. </p>",
      "rawMarkdown": "Hi, if there were such pairs, then you would get a big data leak during training and that would make your validation score unreliable. Hence if that was the case, I would recommend deleting the ones that do have a duplicate in the forged dataset.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3306338,
      "author_name": "sacuscreed",
      "author_url": "",
      "post_date": "10/24/2025 10:45:46",
      "content": "<p>That's what your model has to predict. And mask for the ones your model determine are forged.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3307014,
      "author_name": "michaelhyxie",
      "author_url": "",
      "post_date": "10/26/2025 00:28:18",
      "content": "<p>any submit code available ?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3308088,
      "author_name": "romanfpv",
      "author_url": "",
      "post_date": "10/28/2025 14:36:32",
      "content": "<p>Hi, if there were such pairs, then you would get a big data leak during training and that would make your validation score unreliable. Hence if that was the case, I would recommend deleting the ones that do have a duplicate in the forged dataset. </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3306221": "Hi, i whish to know if there are always  pairs of images (authentic and forged)  that could be shufffled in the test set?",
    "3306338": "That's what your model has to predict. And mask for the ones your model determine are forged.",
    "3307014": "any submit code available ?",
    "3308088": "Hi, if there were such pairs, then you would get a big data leak during training and that would make your validation score unreliable. Hence if that was the case, I would recommend deleting the ones that do have a duplicate in the forged dataset."
  },
  "source": "meta"
}