{
  "id": 462651,
  "title": "Save up inference time predicting only private data",
  "url": "/competitions/blood-vessel-segmentation/discussion/462651",
  "author_name": "",
  "post_date": "2023-12-21T03:01:24.870628900Z",
  "votes": 7,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Since we know there are only 2 kidneys in test, <br>\nwe can easily probe which one is public and which one is private</p>\n<p>The private test uses 30%~ of the data, <br>\nwhich means if we infer only private we have 3x inference time.</p>\n<p>I would rather avoid doing something like that, since these subs would get 0 score on public LB, <br>\nbut can be an advantage for private scoring, how should we handle this?</p>",
  "messages": [
    {
      "id": "2569009",
      "postDate": "12/21/2023 03:01:24",
      "content": "<p>Since we know there are only 2 kidneys in test, <br>\nwe can easily probe which one is public and which one is private</p>\n<p>The private test uses 30%~ of the data, <br>\nwhich means if we infer only private we have 3x inference time.</p>\n<p>I would rather avoid doing something like that, since these subs would get 0 score on public LB, <br>\nbut can be an advantage for private scoring, how should we handle this?</p>",
      "rawMarkdown": "Since we know there are only 2 kidneys in test, \nwe can easily probe which one is public and which one is private\n\nThe private test uses 30%~ of the data, \nwhich means if we infer only private we have 3x inference time.\n\nI would rather avoid doing something like that, since these subs would get 0 score on public LB, \nbut can be an advantage for private scoring, how should we handle this?",
      "votes": null
    },
    {
      "id": "2569023",
      "postDate": "12/21/2023 03:15:22",
      "content": "<p>it is better to ask the organizer if such solution is accepted.<br>\nkaggle/host reserved the rights to reject monetary awards even if it is within the top entries.</p>\n<p>while this is not against the rule, but it is an \"exploitation ?\" of the rule \"loophole\" and not the intention of the host for time limit i think</p>",
      "rawMarkdown": "it is better to ask the organizer if such solution is accepted.\nkaggle/host reserved the rights to reject monetary awards even if it is within the top entries.\n\nwhile this is not against the rule, but it is an \"exploitation ?\" of the rule \"loophole\" and not the intention of the host for time limit i think",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2569023,
      "author_name": "hengck23",
      "author_url": "",
      "post_date": "12/21/2023 03:15:22",
      "content": "<p>it is better to ask the organizer if such solution is accepted.<br>\nkaggle/host reserved the rights to reject monetary awards even if it is within the top entries.</p>\n<p>while this is not against the rule, but it is an \"exploitation ?\" of the rule \"loophole\" and not the intention of the host for time limit i think</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2569009": "Since we know there are only 2 kidneys in test, \nwe can easily probe which one is public and which one is private\n\nThe private test uses 30%~ of the data, \nwhich means if we infer only private we have 3x inference time.\n\nI would rather avoid doing something like that, since these subs would get 0 score on public LB, \nbut can be an advantage for private scoring, how should we handle this?",
    "2569023": "it is better to ask the organizer if such solution is accepted.\nkaggle/host reserved the rights to reject monetary awards even if it is within the top entries.\n\nwhile this is not against the rule, but it is an \"exploitation ?\" of the rule \"loophole\" and not the intention of the host for time limit i think"
  },
  "source": "meta"
}