{
  "id": 113847,
  "title": "Commiting taking too much time",
  "url": "/competitions/ashrae-energy-prediction/discussion/113847",
  "author_name": "",
  "post_date": "2019-10-22T15:01:06.635191200Z",
  "votes": null,
  "comment_count": 4,
  "views": 0,
  "content": "<p>So,I am using lightgbm with 4 folds and using all the 4 to make predictions. I started comit that script and its stills not done even after 10 hours. Initislly there were memory errors while fitting, but after using reduce_mem_function, I didn't see any memory errors. Can anyone help me out with this? Also I have closed the tab where commit was run, is there any way to see current progress of my commit? </p>",
  "messages": [
    {
      "id": "654985",
      "postDate": "10/22/2019 15:01:06",
      "content": "<p>So,I am using lightgbm with 4 folds and using all the 4 to make predictions. I started comit that script and its stills not done even after 10 hours. Initislly there were memory errors while fitting, but after using reduce_mem_function, I didn't see any memory errors. Can anyone help me out with this? Also I have closed the tab where commit was run, is there any way to see current progress of my commit? </p>",
      "rawMarkdown": "So,I am using lightgbm with 4 folds and using all the 4 to make predictions. I started comit that script and its stills not done even after 10 hours. Initislly there were memory errors while fitting, but after using reduce_mem_function, I didn't see any memory errors. Can anyone help me out with this? Also I have closed the tab where commit was run, is there any way to see current progress of my commit?",
      "votes": null
    },
    {
      "id": "655034",
      "postDate": "10/22/2019 16:11:43",
      "content": "<p>Kernels timeout after 9hrs, and you cannot recover anything from it if it does.\nMake sure it will finish within the 9hrs limit, or you'll waste your time waiting.\ne.g. you can run only 1 fold while editing it, and check how long it takes (&lt;= 2hrs, including inference in your case)\nYou can use this function below to automate your kernel. I.e. behave differently when in edit or commit mode:\n<code>\ndef is_interactive():\n    ''' Return True if inside a notebook/kernel in Edit mode, or False if committed '''\n    return 'runtime' in get_ipython().config.IPKernelApp.connection_file\n</code>\nHaving said that I have also a kernel that is taking forever now lol\n2hrs in edit mode (all folds) and already going for 4 hrs after committed... :/</p>",
      "rawMarkdown": "Kernels timeout after 9hrs, and you cannot recover anything from it if it does.\nMake sure it will finish within the 9hrs limit, or you'll waste your time waiting.\ne.g. you can run only 1 fold while editing it, and check how long it takes (&lt;= 2hrs, including inference in your case)\nYou can use this function below to automate your kernel. I.e. behave differently when in edit or commit mode:\n```\ndef is_interactive():\n    ''' Return True if inside a notebook/kernel in Edit mode, or False if committed '''\n    return 'runtime' in get_ipython().config.IPKernelApp.connection_file\n```\nHaving said that I have also a kernel that is taking forever now lol\n2hrs in edit mode (all folds) and already going for 4 hrs after committed... :/",
      "votes": null
    },
    {
      "id": "655101",
      "postDate": "10/22/2019 17:32:42",
      "content": "<p><a href=\"/yourssharmaji\">@yourssharmaji</a>  Reduce your number of estimators to around 200 or something. Your model will run in less than an hour. </p>",
      "rawMarkdown": "yourssharmaji  Reduce your number of estimators to around 200 or something. Your model will run in less than an hour.",
      "votes": null
    },
    {
      "id": "655190",
      "postDate": "10/22/2019 19:30:50",
      "content": "<p>«&nbsp;Commiting taking so much time...&nbsp;»</p>\n\n<p>Maybe the reason why we have only two submissions a day...</p>",
      "rawMarkdown": "«&nbsp;Commiting taking so much time...&nbsp;»\n\nMaybe the reason why we have only two submissions a day...",
      "votes": null
    },
    {
      "id": "655194",
      "postDate": "10/22/2019 19:37:45",
      "content": "<p>I don't think so. The IEEE-Fraud detection had a bigger dataset. Took more time to commit. Still it had 5sub/day.</p>",
      "rawMarkdown": "I don't think so. The IEEE-Fraud detection had a bigger dataset. Took more time to commit. Still it had 5sub/day.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 655034,
      "author_name": "hmendonca",
      "author_url": "",
      "post_date": "10/22/2019 16:11:43",
      "content": "<p>Kernels timeout after 9hrs, and you cannot recover anything from it if it does.\nMake sure it will finish within the 9hrs limit, or you'll waste your time waiting.\ne.g. you can run only 1 fold while editing it, and check how long it takes (&lt;= 2hrs, including inference in your case)\nYou can use this function below to automate your kernel. I.e. behave differently when in edit or commit mode:\n<code>\ndef is_interactive():\n    ''' Return True if inside a notebook/kernel in Edit mode, or False if committed '''\n    return 'runtime' in get_ipython().config.IPKernelApp.connection_file\n</code>\nHaving said that I have also a kernel that is taking forever now lol\n2hrs in edit mode (all folds) and already going for 4 hrs after committed... :/</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 655101,
      "author_name": "themonologue",
      "author_url": "",
      "post_date": "10/22/2019 17:32:42",
      "content": "<p><a href=\"/yourssharmaji\">@yourssharmaji</a>  Reduce your number of estimators to around 200 or something. Your model will run in less than an hour. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 655190,
      "author_name": "pfichou",
      "author_url": "",
      "post_date": "10/22/2019 19:30:50",
      "content": "<p>«&nbsp;Commiting taking so much time...&nbsp;»</p>\n\n<p>Maybe the reason why we have only two submissions a day...</p>",
      "votes": null,
      "replies": [
        {
          "id": 655194,
          "author_name": "themonologue",
          "author_url": "",
          "post_date": "10/22/2019 19:37:45",
          "content": "<p>I don't think so. The IEEE-Fraud detection had a bigger dataset. Took more time to commit. Still it had 5sub/day.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "654985": "So,I am using lightgbm with 4 folds and using all the 4 to make predictions. I started comit that script and its stills not done even after 10 hours. Initislly there were memory errors while fitting, but after using reduce_mem_function, I didn't see any memory errors. Can anyone help me out with this? Also I have closed the tab where commit was run, is there any way to see current progress of my commit?",
    "655034": "Kernels timeout after 9hrs, and you cannot recover anything from it if it does.\nMake sure it will finish within the 9hrs limit, or you'll waste your time waiting.\ne.g. you can run only 1 fold while editing it, and check how long it takes (&lt;= 2hrs, including inference in your case)\nYou can use this function below to automate your kernel. I.e. behave differently when in edit or commit mode:\n```\ndef is_interactive():\n    ''' Return True if inside a notebook/kernel in Edit mode, or False if committed '''\n    return 'runtime' in get_ipython().config.IPKernelApp.connection_file\n```\nHaving said that I have also a kernel that is taking forever now lol\n2hrs in edit mode (all folds) and already going for 4 hrs after committed... :/",
    "655101": "yourssharmaji  Reduce your number of estimators to around 200 or something. Your model will run in less than an hour.",
    "655190": "«&nbsp;Commiting taking so much time...&nbsp;»\n\nMaybe the reason why we have only two submissions a day...",
    "655194": "I don't think so. The IEEE-Fraud detection had a bigger dataset. Took more time to commit. Still it had 5sub/day."
  },
  "source": "meta"
}