{
  "id": 574479,
  "title": "TOP1 changed again",
  "url": "/competitions/stanford-rna-3d-folding/discussion/574479",
  "author_name": "",
  "post_date": "2025-04-22T06:07:50.112755700Z",
  "votes": null,
  "comment_count": 10,
  "views": 0,
  "content": "<p>It is surprising that a new team ranked first and made a gap between the second team</p>",
  "messages": [
    {
      "id": "3184469",
      "postDate": "04/22/2025 06:07:50",
      "content": "<p>It is surprising that a new team ranked first and made a gap between the second team</p>",
      "rawMarkdown": "It is surprising that a new team ranked first and made a gap between the second team",
      "votes": null
    },
    {
      "id": "3184494",
      "postDate": "04/22/2025 06:57:29",
      "content": "<p>Don't worry about that, all you need to do is to prepare for future test.</p>",
      "rawMarkdown": "Don't worry about that, all you need to do is to prepare for future test.",
      "votes": null
    },
    {
      "id": "3184703",
      "postDate": "04/22/2025 11:45:28",
      "content": "<p>0.563 for #1!</p>",
      "rawMarkdown": "0.563 for #1!",
      "votes": null
    },
    {
      "id": "3184711",
      "postDate": "04/22/2025 11:52:53",
      "content": "<p>Tomorrow the LB will be refreshed as new test set will take over</p>",
      "rawMarkdown": "Tomorrow the LB will be refreshed as new test set will take over",
      "votes": null
    },
    {
      "id": "3188087",
      "postDate": "04/27/2025 03:02:15",
      "content": "<p>congratulate to your new ranking, did you fine-tuned the proteinx?</p>",
      "rawMarkdown": "congratulate to your new ranking, did you fine-tuned the proteinx?",
      "votes": null
    },
    {
      "id": "3188092",
      "postDate": "04/27/2025 03:36:28",
      "content": "<p>Sure! However as I mentioned in other discussion, it can lead to overfit when I test locally.</p>",
      "rawMarkdown": "Sure! However as I mentioned in other discussion, it can lead to overfit when I test locally.",
      "votes": null
    },
    {
      "id": "3188094",
      "postDate": "04/27/2025 03:38:58",
      "content": "<p>Yes, I tried using the CASP16 data to finetuned the proteinx  but its score is worse than the unfinetuned version, How did u fix this problem? Did you use extra training data?</p>",
      "rawMarkdown": "Yes, I tried using the CASP16 data to finetuned the proteinx  but its score is worse than the unfinetuned version, How did u fix this problem? Did you use extra training data?",
      "votes": null
    },
    {
      "id": "3188110",
      "postDate": "04/27/2025 04:21:01",
      "content": "<blockquote>\n  <p>Did you use extra training data?</p>\n</blockquote>\n<p>yes, I have several private RNA sequences and their 3D information from the lab where I worked during my graduate studies, not submitted to PDB yet. I'm not sure whether they're reliable.</p>",
      "rawMarkdown": ">Did you use extra training data?\n\nyes, I have several private RNA sequences and their 3D information from the lab where I worked during my graduate studies, not submitted to PDB yet. I'm not sure whether they're reliable.",
      "votes": null
    },
    {
      "id": "3189987",
      "postDate": "04/30/2025 02:43:02",
      "content": "<p>Hi Timmy, can you share a little experience about fine-tuning the proteinx besides using the extra lab data?</p>",
      "rawMarkdown": "Hi Timmy, can you share a little experience about fine-tuning the proteinx besides using the extra lab data?",
      "votes": null
    },
    {
      "id": "3190027",
      "postDate": "04/30/2025 03:59:29",
      "content": "<p>As for finetuning , to be honest, there's no special method, just some conventional optimization methods. More importantly, I employed some additional modeling strategies.</p>",
      "rawMarkdown": "As for finetuning , to be honest, there's no special method, just some conventional optimization methods. More importantly, I employed some additional modeling strategies.",
      "votes": null
    },
    {
      "id": "3190101",
      "postDate": "04/30/2025 05:45:25",
      "content": "<p>Thanks for your reply. Modeling strategies? Is it like the VFOLD pipeline? </p>",
      "rawMarkdown": "Thanks for your reply. Modeling strategies? Is it like the VFOLD pipeline?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3184494,
      "author_name": "sweetyheehee",
      "author_url": "",
      "post_date": "04/22/2025 06:57:29",
      "content": "<p>Don't worry about that, all you need to do is to prepare for future test.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3188087,
          "author_name": "akomwins",
          "author_url": "",
          "post_date": "04/27/2025 03:02:15",
          "content": "<p>congratulate to your new ranking, did you fine-tuned the proteinx?</p>",
          "votes": null,
          "replies": [
            {
              "id": 3188092,
              "author_name": "sweetyheehee",
              "author_url": "",
              "post_date": "04/27/2025 03:36:28",
              "content": "<p>Sure! However as I mentioned in other discussion, it can lead to overfit when I test locally.</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3188094,
                  "author_name": "akomwins",
                  "author_url": "",
                  "post_date": "04/27/2025 03:38:58",
                  "content": "<p>Yes, I tried using the CASP16 data to finetuned the proteinx  but its score is worse than the unfinetuned version, How did u fix this problem? Did you use extra training data?</p>",
                  "votes": null,
                  "replies": [
                    {
                      "id": 3188110,
                      "author_name": "sweetyheehee",
                      "author_url": "",
                      "post_date": "04/27/2025 04:21:01",
                      "content": "<blockquote>\n  <p>Did you use extra training data?</p>\n</blockquote>\n<p>yes, I have several private RNA sequences and their 3D information from the lab where I worked during my graduate studies, not submitted to PDB yet. I'm not sure whether they're reliable.</p>",
                      "votes": null,
                      "replies": [
                        {
                          "id": 3189987,
                          "author_name": "akomwins",
                          "author_url": "",
                          "post_date": "04/30/2025 02:43:02",
                          "content": "<p>Hi Timmy, can you share a little experience about fine-tuning the proteinx besides using the extra lab data?</p>",
                          "votes": null,
                          "replies": [
                            {
                              "id": 3190027,
                              "author_name": "sweetyheehee",
                              "author_url": "",
                              "post_date": "04/30/2025 03:59:29",
                              "content": "<p>As for finetuning , to be honest, there's no special method, just some conventional optimization methods. More importantly, I employed some additional modeling strategies.</p>",
                              "votes": null,
                              "replies": [
                                {
                                  "id": 3190101,
                                  "author_name": "akomwins",
                                  "author_url": "",
                                  "post_date": "04/30/2025 05:45:25",
                                  "content": "<p>Thanks for your reply. Modeling strategies? Is it like the VFOLD pipeline? </p>",
                                  "votes": null,
                                  "replies": []
                                }
                              ]
                            }
                          ]
                        }
                      ]
                    }
                  ]
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": 3184703,
      "author_name": "sangleevxh",
      "author_url": "",
      "post_date": "04/22/2025 11:45:28",
      "content": "<p>0.563 for #1!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3184711,
      "author_name": "arunodhayan",
      "author_url": "",
      "post_date": "04/22/2025 11:52:53",
      "content": "<p>Tomorrow the LB will be refreshed as new test set will take over</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3184469": "It is surprising that a new team ranked first and made a gap between the second team",
    "3184494": "Don't worry about that, all you need to do is to prepare for future test.",
    "3184703": "0.563 for #1!",
    "3184711": "Tomorrow the LB will be refreshed as new test set will take over",
    "3188087": "congratulate to your new ranking, did you fine-tuned the proteinx?",
    "3188092": "Sure! However as I mentioned in other discussion, it can lead to overfit when I test locally.",
    "3188094": "Yes, I tried using the CASP16 data to finetuned the proteinx  but its score is worse than the unfinetuned version, How did u fix this problem? Did you use extra training data?",
    "3188110": ">Did you use extra training data?\n\nyes, I have several private RNA sequences and their 3D information from the lab where I worked during my graduate studies, not submitted to PDB yet. I'm not sure whether they're reliable.",
    "3189987": "Hi Timmy, can you share a little experience about fine-tuning the proteinx besides using the extra lab data?",
    "3190027": "As for finetuning , to be honest, there's no special method, just some conventional optimization methods. More importantly, I employed some additional modeling strategies.",
    "3190101": "Thanks for your reply. Modeling strategies? Is it like the VFOLD pipeline?"
  },
  "source": "meta"
}