{
  "id": 324465,
  "title": "basic question from a beginner",
  "url": "/competitions/sorghum-id-fgvc-9/discussion/324465",
  "author_name": "wbti",
  "post_date": "2022-05-11T21:36:01.716000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Just curious- how biased do we think the classification accuracy at the top of the leaderboard is? ~100 shots at the test set (although \"test set\" seems like the wrong phrase) seems like a lot, but I'm curious if more experienced people have a sense of how much over-fitting can be done over that many iterations. I am a late comer so I'm contemplating how much less information I would have than people submitting predictions for weeks/months.</p>",
  "messages": [
    {
      "id": 1785165,
      "postDate": "2022-05-11T21:36:01.717Z",
      "content": "<p>Just curious- how biased do we think the classification accuracy at the top of the leaderboard is? ~100 shots at the test set (although \"test set\" seems like the wrong phrase) seems like a lot, but I'm curious if more experienced people have a sense of how much over-fitting can be done over that many iterations. I am a late comer so I'm contemplating how much less information I would have than people submitting predictions for weeks/months.</p>",
      "rawMarkdown": "Just curious- how biased do we think the classification accuracy at the top of the leaderboard is? ~100 shots at the test set (although \"test set\" seems like the wrong phrase) seems like a lot, but I'm curious if more experienced people have a sense of how much over-fitting can be done over that many iterations. I am a late comer so I'm contemplating how much less information I would have than people submitting predictions for weeks/months."
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1785165": "Just curious- how biased do we think the classification accuracy at the top of the leaderboard is? ~100 shots at the test set (although \"test set\" seems like the wrong phrase) seems like a lot, but I'm curious if more experienced people have a sense of how much over-fitting can be done over that many iterations. I am a late comer so I'm contemplating how much less information I would have than people submitting predictions for weeks/months."
  }
}