{
  "id": 460956,
  "title": "8 Place Solution(ONODERA part)",
  "url": "/competitions/stanford-ribonanza-rna-folding/discussion/460956",
  "author_name": "",
  "post_date": "2023-12-12T00:46:22.988801200Z",
  "votes": 8,
  "comment_count": 1,
  "views": 0,
  "content": "<p>First of all, thank you for organizing this competition.<br>\nI will put what I did in this competition precisely.</p>\n<h3>Preparing Inputs</h3>\n<p>We generating a lot of inputs; eternafold, rnasoft, rnastructure etc…<br>\nBut none of these can beat eternafold.</p>\n<h3>Model Pipeline</h3>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F317344%2Fdb1a7469b63bcb89587c63d5a74c789c%2F2023-12-12%209.20.39.png?generation=1702340458537215&amp;alt=media\"></p>\n<h3>Blend</h3>\n<p>We tried stacking first, but  it's not as good as nelder-mead. I think it's because competition metric and loss is smaller than Covid Vaccine Degradation comp.</p>\n<h3>Pseudo Label</h3>\n<p>We got a lot of gain from PL.<br>\nKF already described that <a href=\"https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/discussion/460222\" target=\"_blank\">here</a></p>",
  "messages": [
    {
      "id": "2558166",
      "postDate": "12/12/2023 00:46:22",
      "content": "<p>First of all, thank you for organizing this competition.<br>\nI will put what I did in this competition precisely.</p>\n<h3>Preparing Inputs</h3>\n<p>We generating a lot of inputs; eternafold, rnasoft, rnastructure etc…<br>\nBut none of these can beat eternafold.</p>\n<h3>Model Pipeline</h3>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F317344%2Fdb1a7469b63bcb89587c63d5a74c789c%2F2023-12-12%209.20.39.png?generation=1702340458537215&amp;alt=media\"></p>\n<h3>Blend</h3>\n<p>We tried stacking first, but  it's not as good as nelder-mead. I think it's because competition metric and loss is smaller than Covid Vaccine Degradation comp.</p>\n<h3>Pseudo Label</h3>\n<p>We got a lot of gain from PL.<br>\nKF already described that <a href=\"https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/discussion/460222\" target=\"_blank\">here</a></p>",
      "rawMarkdown": "First of all, thank you for organizing this competition.\nI will put what I did in this competition precisely.\n\n### Preparing Inputs\nWe generating a lot of inputs; eternafold, rnasoft, rnastructure etc...\nBut none of these can beat eternafold.\n\n### Model Pipeline\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F317344%2Fdb1a7469b63bcb89587c63d5a74c789c%2F2023-12-12%209.20.39.png?generation=1702340458537215&alt=media\" width=\"50%\">\n\n### Blend\nWe tried stacking first, but  it's not as good as nelder-mead. I think it's because competition metric and loss is smaller than Covid Vaccine Degradation comp.\n\n### Pseudo Label\nWe got a lot of gain from PL.\nKF already described that [here](https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/discussion/460222)",
      "votes": null
    },
    {
      "id": "2560250",
      "postDate": "12/13/2023 11:55:52",
      "content": "<p><a href=\"https://www.kaggle.com/onodera\" target=\"_blank\">@onodera</a> interesting model pipeline, thanks for sharing</p>",
      "rawMarkdown": "onodera interesting model pipeline, thanks for sharing",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2560250,
      "author_name": "volodymyrpivoshenko",
      "author_url": "",
      "post_date": "12/13/2023 11:55:52",
      "content": "<p><a href=\"https://www.kaggle.com/onodera\" target=\"_blank\">@onodera</a> interesting model pipeline, thanks for sharing</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2558166": "First of all, thank you for organizing this competition.\nI will put what I did in this competition precisely.\n\n### Preparing Inputs\nWe generating a lot of inputs; eternafold, rnasoft, rnastructure etc...\nBut none of these can beat eternafold.\n\n### Model Pipeline\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F317344%2Fdb1a7469b63bcb89587c63d5a74c789c%2F2023-12-12%209.20.39.png?generation=1702340458537215&alt=media\" width=\"50%\">\n\n### Blend\nWe tried stacking first, but  it's not as good as nelder-mead. I think it's because competition metric and loss is smaller than Covid Vaccine Degradation comp.\n\n### Pseudo Label\nWe got a lot of gain from PL.\nKF already described that [here](https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/discussion/460222)",
    "2560250": "onodera interesting model pipeline, thanks for sharing"
  },
  "source": "meta"
}