{
  "id": 663917,
  "title": "Any luck with classification models?",
  "url": "/competitions/recodai-luc-scientific-image-forgery-detection/discussion/663917",
  "author_name": "",
  "post_date": "2025-12-21T13:15:22.771802Z",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I just started this competition and wanted to explore classification -&gt; attention maps approach since it might capture copy-paste forgeries inherently, but it seems like my models are not learning anything at all. Log loss is stuck at 0.69 and roc auc is stuck at 0.5. I'm not sure if there's a bug in my pipeline or is it like this for everyone?</p>",
  "messages": [
    {
      "id": "3380019",
      "postDate": "12/21/2025 13:15:22",
      "content": "<p>I just started this competition and wanted to explore classification -&gt; attention maps approach since it might capture copy-paste forgeries inherently, but it seems like my models are not learning anything at all. Log loss is stuck at 0.69 and roc auc is stuck at 0.5. I'm not sure if there's a bug in my pipeline or is it like this for everyone?</p>",
      "rawMarkdown": "I just started this competition and wanted to explore classification -> attention maps approach since it might capture copy-paste forgeries inherently, but it seems like my models are not learning anything at all. Log loss is stuck at 0.69 and roc auc is stuck at 0.5. I'm not sure if there's a bug in my pipeline or is it like this for everyone?",
      "votes": null
    },
    {
      "id": "3380229",
      "postDate": "12/22/2025 00:59:13",
      "content": "<p>Same with me; i can't seem to get past 0.69 for the classification tasks; 0.62~0.63 seems to be the average </p>",
      "rawMarkdown": "Same with me; i can't seem to get past 0.69 for the classification tasks; 0.62~0.63 seems to be the average",
      "votes": null
    },
    {
      "id": "3380264",
      "postDate": "12/22/2025 05:05:21",
      "content": "<p>Thanks for the information. What do you mean by 0.62-0.63 is the average? Is it based on the papers?</p>",
      "rawMarkdown": "Thanks for the information. What do you mean by 0.62-0.63 is the average? Is it based on the papers?",
      "votes": null
    },
    {
      "id": "3380338",
      "postDate": "12/22/2025 09:02:21",
      "content": "<p>it's the average confidence score for the classification predictions</p>",
      "rawMarkdown": "it's the average confidence score for the classification predictions",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3380229,
      "author_name": "samu2505",
      "author_url": "",
      "post_date": "12/22/2025 00:59:13",
      "content": "<p>Same with me; i can't seem to get past 0.69 for the classification tasks; 0.62~0.63 seems to be the average </p>",
      "votes": null,
      "replies": [
        {
          "id": 3380264,
          "author_name": "gunesevitan",
          "author_url": "",
          "post_date": "12/22/2025 05:05:21",
          "content": "<p>Thanks for the information. What do you mean by 0.62-0.63 is the average? Is it based on the papers?</p>",
          "votes": null,
          "replies": [
            {
              "id": 3380338,
              "author_name": "samu2505",
              "author_url": "",
              "post_date": "12/22/2025 09:02:21",
              "content": "<p>it's the average confidence score for the classification predictions</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3380019": "I just started this competition and wanted to explore classification -> attention maps approach since it might capture copy-paste forgeries inherently, but it seems like my models are not learning anything at all. Log loss is stuck at 0.69 and roc auc is stuck at 0.5. I'm not sure if there's a bug in my pipeline or is it like this for everyone?",
    "3380229": "Same with me; i can't seem to get past 0.69 for the classification tasks; 0.62~0.63 seems to be the average",
    "3380264": "Thanks for the information. What do you mean by 0.62-0.63 is the average? Is it based on the papers?",
    "3380338": "it's the average confidence score for the classification predictions"
  },
  "source": "meta"
}