{
  "id": 250705,
  "title": "Quick versioning makes Notebook Threw Exception",
  "url": "/competitions/mlb-player-digital-engagement-forecasting/discussion/250705",
  "author_name": "",
  "post_date": "2021-07-04T01:49:11.445138200Z",
  "votes": 3,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Finally, I get <strong>Notebook Threw Exception</strong> on submission.<br>\nUntil now, I believe that the errors of other discussions were caused by others faulty assumptions, but that was an arrogant idea. Let me begin by saying that I am really sorry for having such thoughts.</p>\n<p>my situation:</p>\n<ol>\n<li>data pipeline is same with one of my <strong>successes</strong> kernel.</li>\n<li>submitted kernel was save by quick versioning.</li>\n<li>after execute, just before Quick version, I change only number of epochs </li>\n<li>timing of throwing error when <strong>data preprocessing</strong> or <strong>print epochs</strong></li>\n</ol>\n<p>also, this kernel was execute without error on commit mode.<br>\nas I describe above, pipeline will not have error</p>\n<p>then suspicion was,</p>\n<ol>\n<li>testing environment do not allow <strong>!mkdir</strong>  or  <strong>import time</strong></li>\n<li>testing environment input data changed without notice ( such as they drop some cols in players.csv, or no players.csv file in input )</li>\n<li>while scoring, they do not allow not equal output of submission ( in other words, do not allow if submitted outputs != executed output) </li>\n<li>some libraries(such as torch, tensorflow, pandas, numpy, LightGBM …) versions are not equal between testing env and our commit saving env</li>\n<li>any variable name do not allowed?</li>\n<li>memory error? possible. but, error type is not <strong>Notebook Exceeded Allowed Compute</strong></li>\n<li>some libraries are not in testing env which are in commit mode?</li>\n</ol>\n<p>still I don't know actual reason,<br>\nBut <strong>Committed (not Quick saved) kernel do not throw error.</strong><br>\nThis is not stated in overview of the competition.</p>\n<p>then remaining possibilities are,</p>\n<ol>\n<li>while scoring, they do not allow unequal output of submission ( in other words, do not allow if submitted kernel's outputs != executed output) </li>\n<li>they do not allow 'Quick version'</li>\n</ol>\n<p>In any case, I have to spend double time resources. <br>\none for commit, one for scoring<br>\nBut time is gold ……..</p>\n<p>It do not means Quick version makes exception always, but it might effect to some cases (include my case) </p>",
  "messages": [
    {
      "id": "1375195",
      "postDate": "07/04/2021 01:49:11",
      "content": "<p>Finally, I get <strong>Notebook Threw Exception</strong> on submission.<br>\nUntil now, I believe that the errors of other discussions were caused by others faulty assumptions, but that was an arrogant idea. Let me begin by saying that I am really sorry for having such thoughts.</p>\n<p>my situation:</p>\n<ol>\n<li>data pipeline is same with one of my <strong>successes</strong> kernel.</li>\n<li>submitted kernel was save by quick versioning.</li>\n<li>after execute, just before Quick version, I change only number of epochs </li>\n<li>timing of throwing error when <strong>data preprocessing</strong> or <strong>print epochs</strong></li>\n</ol>\n<p>also, this kernel was execute without error on commit mode.<br>\nas I describe above, pipeline will not have error</p>\n<p>then suspicion was,</p>\n<ol>\n<li>testing environment do not allow <strong>!mkdir</strong>  or  <strong>import time</strong></li>\n<li>testing environment input data changed without notice ( such as they drop some cols in players.csv, or no players.csv file in input )</li>\n<li>while scoring, they do not allow not equal output of submission ( in other words, do not allow if submitted outputs != executed output) </li>\n<li>some libraries(such as torch, tensorflow, pandas, numpy, LightGBM …) versions are not equal between testing env and our commit saving env</li>\n<li>any variable name do not allowed?</li>\n<li>memory error? possible. but, error type is not <strong>Notebook Exceeded Allowed Compute</strong></li>\n<li>some libraries are not in testing env which are in commit mode?</li>\n</ol>\n<p>still I don't know actual reason,<br>\nBut <strong>Committed (not Quick saved) kernel do not throw error.</strong><br>\nThis is not stated in overview of the competition.</p>\n<p>then remaining possibilities are,</p>\n<ol>\n<li>while scoring, they do not allow unequal output of submission ( in other words, do not allow if submitted kernel's outputs != executed output) </li>\n<li>they do not allow 'Quick version'</li>\n</ol>\n<p>In any case, I have to spend double time resources. <br>\none for commit, one for scoring<br>\nBut time is gold ……..</p>\n<p>It do not means Quick version makes exception always, but it might effect to some cases (include my case) </p>",
      "rawMarkdown": "Finally, I get **Notebook Threw Exception** on submission.\nUntil now, I believe that the errors of other discussions were caused by others faulty assumptions, but that was an arrogant idea. Let me begin by saying that I am really sorry for having such thoughts.\n\nmy situation:\n1. data pipeline is same with one of my **successes** kernel.\n2. submitted kernel was save by quick versioning.\n3. after execute, just before Quick version, I change only number of epochs \n4. timing of throwing error when **data preprocessing** or **print epochs**\n\nalso, this kernel was execute without error on commit mode.\nas I describe above, pipeline will not have error\n\nthen suspicion was,\n1. testing environment do not allow **!mkdir**  or  **import time**\n2. testing environment input data changed without notice ( such as they drop some cols in players.csv, or no players.csv file in input )\n3. while scoring, they do not allow not equal output of submission ( in other words, do not allow if submitted outputs != executed output) \n4. some libraries(such as torch, tensorflow, pandas, numpy, LightGBM ...) versions are not equal between testing env and our commit saving env\n5. any variable name do not allowed?\n6. memory error? possible. but, error type is not **Notebook Exceeded Allowed Compute**\n7. some libraries are not in testing env which are in commit mode?\n\n\nstill I don't know actual reason,\nBut **Committed (not Quick saved) kernel do not throw error.**\nThis is not stated in overview of the competition.\n\n\nthen remaining possibilities are,\n1. while scoring, they do not allow unequal output of submission ( in other words, do not allow if submitted kernel's outputs != executed output) \n2. they do not allow 'Quick version'\n\n\nIn any case, I have to spend double time resources. \none for commit, one for scoring\nBut time is gold ........\n\nIt do not means Quick version makes exception always, but it might effect to some cases (include my case)",
      "votes": null
    },
    {
      "id": "1380503",
      "postDate": "07/08/2021 06:16:55",
      "content": "<p>I am receiving the same error.  I am submitting and within 60 seconds I get the error. It only takes a minute to read the csv.  I have followed every steps you mentioned above.Any suggestions will help.!!</p>",
      "rawMarkdown": "I am receiving the same error.  I am submitting and within 60 seconds I get the error. It only takes a minute to read the csv.  I have followed every steps you mentioned above.Any suggestions will help.!!",
      "votes": null
    },
    {
      "id": "1382519",
      "postDate": "07/10/2021 01:33:52",
      "content": "<p>First you need to find out which line is causing the error. You need to submit multiple time. I recommend add single time.sleep(120) in your code and check when it stop, 60 sec or 180 sec.<br>\nNext write the input specification you assume on that line<br>\nFinally, it should be compared against the actual input statement to derive possible errors.</p>",
      "rawMarkdown": "First you need to find out which line is causing the error. You need to submit multiple time. I recommend add single time.sleep(120) in your code and check when it stop, 60 sec or 180 sec.\nNext write the input specification you assume on that line\nFinally, it should be compared against the actual input statement to derive possible errors.",
      "votes": null
    },
    {
      "id": "1382591",
      "postDate": "07/10/2021 04:08:28",
      "content": "<p>In some past competitions the order of the player/date would have been important - sorting before submission killed things.</p>\n<p>Can anyone confirm that the order of the 1187 players/dates is not critical to a successful submission?</p>",
      "rawMarkdown": "In some past competitions the order of the player/date would have been important - sorting before submission killed things.\n\nCan anyone confirm that the order of the 1187 players/dates is not critical to a successful submission?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1380503,
      "author_name": "debojit23",
      "author_url": "",
      "post_date": "07/08/2021 06:16:55",
      "content": "<p>I am receiving the same error.  I am submitting and within 60 seconds I get the error. It only takes a minute to read the csv.  I have followed every steps you mentioned above.Any suggestions will help.!!</p>",
      "votes": null,
      "replies": [
        {
          "id": 1382519,
          "author_name": "assign",
          "author_url": "",
          "post_date": "07/10/2021 01:33:52",
          "content": "<p>First you need to find out which line is causing the error. You need to submit multiple time. I recommend add single time.sleep(120) in your code and check when it stop, 60 sec or 180 sec.<br>\nNext write the input specification you assume on that line<br>\nFinally, it should be compared against the actual input statement to derive possible errors.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1382591,
      "author_name": "pcjimmmy",
      "author_url": "",
      "post_date": "07/10/2021 04:08:28",
      "content": "<p>In some past competitions the order of the player/date would have been important - sorting before submission killed things.</p>\n<p>Can anyone confirm that the order of the 1187 players/dates is not critical to a successful submission?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1375195": "Finally, I get **Notebook Threw Exception** on submission.\nUntil now, I believe that the errors of other discussions were caused by others faulty assumptions, but that was an arrogant idea. Let me begin by saying that I am really sorry for having such thoughts.\n\nmy situation:\n1. data pipeline is same with one of my **successes** kernel.\n2. submitted kernel was save by quick versioning.\n3. after execute, just before Quick version, I change only number of epochs \n4. timing of throwing error when **data preprocessing** or **print epochs**\n\nalso, this kernel was execute without error on commit mode.\nas I describe above, pipeline will not have error\n\nthen suspicion was,\n1. testing environment do not allow **!mkdir**  or  **import time**\n2. testing environment input data changed without notice ( such as they drop some cols in players.csv, or no players.csv file in input )\n3. while scoring, they do not allow not equal output of submission ( in other words, do not allow if submitted outputs != executed output) \n4. some libraries(such as torch, tensorflow, pandas, numpy, LightGBM ...) versions are not equal between testing env and our commit saving env\n5. any variable name do not allowed?\n6. memory error? possible. but, error type is not **Notebook Exceeded Allowed Compute**\n7. some libraries are not in testing env which are in commit mode?\n\n\nstill I don't know actual reason,\nBut **Committed (not Quick saved) kernel do not throw error.**\nThis is not stated in overview of the competition.\n\n\nthen remaining possibilities are,\n1. while scoring, they do not allow unequal output of submission ( in other words, do not allow if submitted kernel's outputs != executed output) \n2. they do not allow 'Quick version'\n\n\nIn any case, I have to spend double time resources. \none for commit, one for scoring\nBut time is gold ........\n\nIt do not means Quick version makes exception always, but it might effect to some cases (include my case)",
    "1380503": "I am receiving the same error.  I am submitting and within 60 seconds I get the error. It only takes a minute to read the csv.  I have followed every steps you mentioned above.Any suggestions will help.!!",
    "1382519": "First you need to find out which line is causing the error. You need to submit multiple time. I recommend add single time.sleep(120) in your code and check when it stop, 60 sec or 180 sec.\nNext write the input specification you assume on that line\nFinally, it should be compared against the actual input statement to derive possible errors.",
    "1382591": "In some past competitions the order of the player/date would have been important - sorting before submission killed things.\n\nCan anyone confirm that the order of the 1187 players/dates is not critical to a successful submission?"
  },
  "source": "meta"
}