{
  "id": 171263,
  "title": "Can we submit two or more at the same time (with pseudo labeling) ? ",
  "url": "/competitions/global-wheat-detection/discussion/171263",
  "author_name": "Marco Wu",
  "post_date": "2020-07-31T05:39:22.318000",
  "votes": 1,
  "comment_count": 3,
  "views": 0,
  "content": "<p>If I submit two or more pseudo labeling notebook at the same time, would it influences gpu run time ? because I submitting two same notebook simultaneously, only one could get score,and the other would get notebook timeout(same time run two different notebook with pseudo labeling).  </p>",
  "messages": [
    {
      "id": 952644,
      "postDate": "2020-07-31T05:39:22.320Z",
      "content": "<p>If I submit two or more pseudo labeling notebook at the same time, would it influences gpu run time ? because I submitting two same notebook simultaneously, only one could get score,and the other would get notebook timeout(same time run two different notebook with pseudo labeling).  </p>",
      "rawMarkdown": "If I submit two or more pseudo labeling notebook at the same time, would it influences gpu run time ? because I submitting two same notebook simultaneously, only one could get score,and the other would get notebook timeout(same time run two different notebook with pseudo labeling).  ",
      "votes": 1
    },
    {
      "id": 952732,
      "postDate": "2020-07-31T07:10:36.350Z",
      "content": "<p>GPU run time should not be affected by the number of submitted notebooks running at the same time. Maybe there are some changes in the second notebook that cause it to run a little longer and pass the time limit?</p>\n\n<p>By the way, you can add time limit condition to prevent kernel timeout in your code something like this:</p>\n\n<p>```</p>\n\n<h1>at the beginning</h1>\n\n<p>TIME_LIMIT = 8.9*60*60 # about 9 hours runtime limit\nKERNEL_START_TIME = time.time()</p>\n\n<h1>...</h1>\n\n<h1>then at some point</h1>\n\n<p>if time.time() - KERNEL_START_TIME &gt; TIME_LIMIT:</p>\n\n<h1>start exiting process</h1>\n\n<h1>...</h1>\n\n<p>```</p>",
      "rawMarkdown": "GPU run time should not be affected by the number of submitted notebooks running at the same time. Maybe there are some changes in the second notebook that cause it to run a little longer and pass the time limit?\n\nBy the way, you can add time limit condition to prevent kernel timeout in your code something like this:\n\n```\n# at the beginning\nTIME_LIMIT = 8.9*60*60 # about 9 hours runtime limit\nKERNEL_START_TIME = time.time()\n# ...\n# then at some point\nif time.time() - KERNEL_START_TIME &gt; TIME_LIMIT:\n# start exiting process \n# ...\n```",
      "votes": 2,
      "replies": [
        {
          "id": 952789,
          "postDate": "2020-07-31T08:17:37.757Z",
          "content": "<p>Maybe just a coincidence in my case,because my successful notebook only got about 7 hours run times( in same epoch 、same optimizer...). I don't know whether if other reasons about gpu distribution,but I still agree with your opinion that should not affected by the number of submitted notebooks,so I will check more detail contents in my notebook again. And Thanks for your advice, I will add time limit into my notebook!!</p>",
          "rawMarkdown": "Maybe just a coincidence in my case,because my successful notebook only got about 7 hours run times( in same epoch 、same optimizer...). I don't know whether if other reasons about gpu distribution,but I still agree with your opinion that should not affected by the number of submitted notebooks,so I will check more detail contents in my notebook again. And Thanks for your advice, I will add time limit into my notebook!!"
        }
      ]
    },
    {
      "id": 952835,
      "postDate": "2020-07-31T08:59:34.143Z",
      "rawMarkdown": "",
      "votes": -3,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 952732,
      "author_name": "alexx",
      "author_url": "",
      "post_date": "2020-07-31T07:10:36.350000",
      "content": "<p>GPU run time should not be affected by the number of submitted notebooks running at the same time. Maybe there are some changes in the second notebook that cause it to run a little longer and pass the time limit?</p>\n\n<p>By the way, you can add time limit condition to prevent kernel timeout in your code something like this:</p>\n\n<p>```</p>\n\n<h1>at the beginning</h1>\n\n<p>TIME_LIMIT = 8.9*60*60 # about 9 hours runtime limit\nKERNEL_START_TIME = time.time()</p>\n\n<h1>...</h1>\n\n<h1>then at some point</h1>\n\n<p>if time.time() - KERNEL_START_TIME &gt; TIME_LIMIT:</p>\n\n<h1>start exiting process</h1>\n\n<h1>...</h1>\n\n<p>```</p>",
      "votes": 2,
      "replies": [
        {
          "id": 952789,
          "author_name": "Marco Wu",
          "author_url": "",
          "post_date": "2020-07-31T08:17:37.757000",
          "content": "<p>Maybe just a coincidence in my case,because my successful notebook only got about 7 hours run times( in same epoch 、same optimizer...). I don't know whether if other reasons about gpu distribution,but I still agree with your opinion that should not affected by the number of submitted notebooks,so I will check more detail contents in my notebook again. And Thanks for your advice, I will add time limit into my notebook!!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 952835,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-07-31T08:59:34.143000",
      "content": "",
      "votes": -3,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "952644": "If I submit two or more pseudo labeling notebook at the same time, would it influences gpu run time ? because I submitting two same notebook simultaneously, only one could get score,and the other would get notebook timeout(same time run two different notebook with pseudo labeling).  ",
    "952732": "GPU run time should not be affected by the number of submitted notebooks running at the same time. Maybe there are some changes in the second notebook that cause it to run a little longer and pass the time limit?\n\nBy the way, you can add time limit condition to prevent kernel timeout in your code something like this:\n\n```\n# at the beginning\nTIME_LIMIT = 8.9*60*60 # about 9 hours runtime limit\nKERNEL_START_TIME = time.time()\n# ...\n# then at some point\nif time.time() - KERNEL_START_TIME &gt; TIME_LIMIT:\n# start exiting process \n# ...\n```",
    "952835": ""
  }
}