{
  "id": 98553,
  "title": "Submittion is not over ...",
  "url": "/competitions/aptos2019-blindness-detection/discussion/98553",
  "author_name": "kaerururu",
  "post_date": "2019-07-04T15:05:02.481000",
  "votes": 2,
  "comment_count": 7,
  "views": 0,
  "content": "<p>My Kernel took 2000sec to commit  (and successed committing) is still running although it took 3h...  😥</p>\n\n<p>What Happend ?  Are there one in the same problem...</p>",
  "messages": [
    {
      "id": 568233,
      "postDate": "2019-07-04T15:17:28.443Z",
      "content": "<p>Its synchronous kernel competition meaning rerun on full test set on submission. Private test set is around 20GB while public is 1.5 GB, so large inference time is expected.</p>",
      "rawMarkdown": "Its synchronous kernel competition meaning rerun on full test set on submission. Private test set is around 20GB while public is 1.5 GB, so large inference time is expected.",
      "votes": 2
    },
    {
      "id": 568225,
      "postDate": "2019-07-04T15:05:02.480Z",
      "content": "<p>My Kernel took 2000sec to commit  (and successed committing) is still running although it took 3h...  😥</p>\n\n<p>What Happend ?  Are there one in the same problem...</p>",
      "rawMarkdown": "My Kernel took 2000sec to commit  (and successed committing) is still running although it took 3h...  😥\n\nWhat Happend ?  Are there one in the same problem...\n\n",
      "votes": 2
    },
    {
      "id": 569850,
      "postDate": "2019-07-07T12:54:21.473Z",
      "content": "<p><a href=\"/cv13j0\">@cv13j0</a> </p>\n\n<p>Hi! If you take time to commit, Just try split training and inference!\nIn other words,  try below process.</p>\n\n<ol>\n<li>Save your weight in training kernel. </li>\n<li>And upload your weight as a dataset.</li>\n<li>In inference kernel, load your weight and do prediction.</li>\n</ol>\n\n<p>So, you can save your time a little, and use full kernel time (~9h) to inference! </p>",
      "rawMarkdown": "@cv13j0 \n\nHi! If you take time to commit, Just try split training and inference!\nIn other words,  try below process.\n\n1. Save your weight in training kernel. \n2. And upload your weight as a dataset.\n3.  In inference kernel, load your weight and do prediction.\n\nSo, you can save your time a little, and use full kernel time (~9h) to inference! "
    },
    {
      "id": 569114,
      "postDate": "2019-07-06T05:04:04.293Z",
      "content": "<p>Hi! <a href=\"/kaerunantoka\">@kaerunantoka</a> my commits are taking a long time, up to 3 hours, looking for options to speed up things a little bit.</p>",
      "rawMarkdown": "Hi! @kaerunantoka my commits are taking a long time, up to 3 hours, looking for options to speed up things a little bit."
    },
    {
      "id": 568564,
      "postDate": "2019-07-05T05:45:57.420Z",
      "content": "<p>Same problem here... I see that the full test is much bigger (the public one is 15% of total), but this is too slow.\nMy inference kernel took 500 seconds, instead, my submission is still running after 8h!</p>\n\n<p>EDIT: Failed after 10h because it exceeded the time limit... That's weird! I'll try creating a new kernel with the same code..</p>",
      "rawMarkdown": "Same problem here... I see that the full test is much bigger (the public one is 15% of total), but this is too slow.\nMy inference kernel took 500 seconds, instead, my submission is still running after 8h!\n\nEDIT: Failed after 10h because it exceeded the time limit... That's weird! I'll try creating a new kernel with the same code.."
    },
    {
      "id": 568249,
      "postDate": "2019-07-04T15:43:49.367Z",
      "content": "<p><a href=\"/mktripathi456\">@mktripathi456</a> </p>\n\n<p>Thank you very mach ;) !!!</p>\n\n<p>I did not take it into account.</p>",
      "rawMarkdown": "@mktripathi456 \n\nThank you very mach ;) !!!\n\nI did not take it into account."
    },
    {
      "id": 569453,
      "postDate": "2019-07-06T17:04:17.893Z",
      "rawMarkdown": "",
      "isDeleted": true
    },
    {
      "id": 568240,
      "postDate": "2019-07-04T15:27:14.883Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 568233,
      "author_name": "Murlikrishnan Tripathi",
      "author_url": "",
      "post_date": "2019-07-04T15:17:28.443000",
      "content": "<p>Its synchronous kernel competition meaning rerun on full test set on submission. Private test set is around 20GB while public is 1.5 GB, so large inference time is expected.</p>",
      "votes": 2,
      "replies": []
    },
    {
      "id": 569850,
      "author_name": "kaerururu",
      "author_url": "",
      "post_date": "2019-07-07T12:54:21.473000",
      "content": "<p><a href=\"/cv13j0\">@cv13j0</a> </p>\n\n<p>Hi! If you take time to commit, Just try split training and inference!\nIn other words,  try below process.</p>\n\n<ol>\n<li>Save your weight in training kernel. </li>\n<li>And upload your weight as a dataset.</li>\n<li>In inference kernel, load your weight and do prediction.</li>\n</ol>\n\n<p>So, you can save your time a little, and use full kernel time (~9h) to inference! </p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 569114,
      "author_name": "C4rl05/V",
      "author_url": "",
      "post_date": "2019-07-06T05:04:04.293000",
      "content": "<p>Hi! <a href=\"/kaerunantoka\">@kaerunantoka</a> my commits are taking a long time, up to 3 hours, looking for options to speed up things a little bit.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 568564,
      "author_name": "Federico Raimondi Cominesi",
      "author_url": "",
      "post_date": "2019-07-05T05:45:57.420000",
      "content": "<p>Same problem here... I see that the full test is much bigger (the public one is 15% of total), but this is too slow.\nMy inference kernel took 500 seconds, instead, my submission is still running after 8h!</p>\n\n<p>EDIT: Failed after 10h because it exceeded the time limit... That's weird! I'll try creating a new kernel with the same code..</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 568249,
      "author_name": "kaerururu",
      "author_url": "",
      "post_date": "2019-07-04T15:43:49.367000",
      "content": "<p><a href=\"/mktripathi456\">@mktripathi456</a> </p>\n\n<p>Thank you very mach ;) !!!</p>\n\n<p>I did not take it into account.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 569453,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-07-06T17:04:17.893000",
      "content": "",
      "votes": 0,
      "replies": []
    },
    {
      "id": 568240,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-07-04T15:27:14.883000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "568233": "Its synchronous kernel competition meaning rerun on full test set on submission. Private test set is around 20GB while public is 1.5 GB, so large inference time is expected.",
    "568225": "My Kernel took 2000sec to commit  (and successed committing) is still running although it took 3h...  😥\n\nWhat Happend ?  Are there one in the same problem...\n\n",
    "569850": "@cv13j0 \n\nHi! If you take time to commit, Just try split training and inference!\nIn other words,  try below process.\n\n1. Save your weight in training kernel. \n2. And upload your weight as a dataset.\n3.  In inference kernel, load your weight and do prediction.\n\nSo, you can save your time a little, and use full kernel time (~9h) to inference! ",
    "569114": "Hi! @kaerunantoka my commits are taking a long time, up to 3 hours, looking for options to speed up things a little bit.",
    "568564": "Same problem here... I see that the full test is much bigger (the public one is 15% of total), but this is too slow.\nMy inference kernel took 500 seconds, instead, my submission is still running after 8h!\n\nEDIT: Failed after 10h because it exceeded the time limit... That's weird! I'll try creating a new kernel with the same code..",
    "568249": "@mktripathi456 \n\nThank you very mach ;) !!!\n\nI did not take it into account.",
    "569453": "",
    "568240": ""
  }
}