{
  "id": 682866,
  "title": "are we have to use ML models ?",
  "url": "/competitions/stanford-rna-3d-folding-2/discussion/682866",
  "author_name": "AhmedMHussien",
  "post_date": "2026-03-19T06:56:31.633000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>my pipeline is largely heuristic/non-ML (Vienna + greedy consensus + hand-built geometry) - while not getting more score than 0.137 - i found that code shared on site uses (hybrid TBM + Protenix deep model) and get a score of 0.428, actually i have compared my code (non ML based) ) on <a href=\"https://molstar.org/viewer/\" target=\"_blank\">https://molstar.org/viewer/</a> with the original known to humanity shape and i found it identical - so why i getting a lower score / do i have to use ML models ??! </p>",
  "messages": [
    {
      "id": 3423949,
      "postDate": "2026-03-19T06:56:31.633Z",
      "content": "<p>my pipeline is largely heuristic/non-ML (Vienna + greedy consensus + hand-built geometry) - while not getting more score than 0.137 - i found that code shared on site uses (hybrid TBM + Protenix deep model) and get a score of 0.428, actually i have compared my code (non ML based) ) on <a href=\"https://molstar.org/viewer/\" target=\"_blank\">https://molstar.org/viewer/</a> with the original known to humanity shape and i found it identical - so why i getting a lower score / do i have to use ML models ??! </p>",
      "rawMarkdown": "my pipeline is largely heuristic/non-ML (Vienna + greedy consensus + hand-built geometry) - while not getting more score than 0.137 - i found that code shared on site uses (hybrid TBM + Protenix deep model) and get a score of 0.428, actually i have compared my code (non ML based) ) on https://molstar.org/viewer/ with the original known to humanity shape and i found it identical - so why i getting a lower score / do i have to use ML models ??! "
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "3423949": "my pipeline is largely heuristic/non-ML (Vienna + greedy consensus + hand-built geometry) - while not getting more score than 0.137 - i found that code shared on site uses (hybrid TBM + Protenix deep model) and get a score of 0.428, actually i have compared my code (non ML based) ) on https://molstar.org/viewer/ with the original known to humanity shape and i found it identical - so why i getting a lower score / do i have to use ML models ??! "
  }
}