{
  "id": 209631,
  "title": "notebook submission running into timeout error.",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/209631",
  "author_name": "",
  "post_date": "2021-01-08T04:20:07.279494200Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>I am not able to submit my work. I have tried multiple times but notebook is running into timeout issue. Can someone help me here? Thanks in advance. </p>\n<p>I am attaching the screen shot.</p>",
  "messages": [
    {
      "id": "1143791",
      "postDate": "01/08/2021 04:20:07",
      "content": "<p>I am not able to submit my work. I have tried multiple times but notebook is running into timeout issue. Can someone help me here? Thanks in advance. </p>\n<p>I am attaching the screen shot.</p>",
      "rawMarkdown": "I am not able to submit my work. I have tried multiple times but notebook is running into timeout issue. Can someone help me here? Thanks in advance. \n\nI am attaching the screen shot.",
      "votes": null
    },
    {
      "id": "1143795",
      "postDate": "01/08/2021 04:24:02",
      "content": "<p>If its a timeout issue, then probably your notebook is running for more than 9 hours at a time which is not allowed as the session limit is 9 hours. If you want to do that, kindly divide it into two parts. 1st is Training and 2nd is Inferring. Do the 1st part in multiple sessions and upload the trained model in some private dataset. Once you get that done, you can use that pretrained model from there in your inference notebook which won't timeout. </p>",
      "rawMarkdown": "If its a timeout issue, then probably your notebook is running for more than 9 hours at a time which is not allowed as the session limit is 9 hours. If you want to do that, kindly divide it into two parts. 1st is Training and 2nd is Inferring. Do the 1st part in multiple sessions and upload the trained model in some private dataset. Once you get that done, you can use that pretrained model from there in your inference notebook which won't timeout.",
      "votes": null
    },
    {
      "id": "1144153",
      "postDate": "01/08/2021 09:11:50",
      "content": "<p>You can divide your notebook two parts. First, preprocessing and modelling. After you save your model, you can build submission notebook and use model that you saved. You can build your private dataset that save your own models. </p>",
      "rawMarkdown": "You can divide your notebook two parts. First, preprocessing and modelling. After you save your model, you can build submission notebook and use model that you saved. You can build your private dataset that save your own models.",
      "votes": null
    },
    {
      "id": "1144481",
      "postDate": "01/08/2021 13:36:16",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/mynkdsi1011\" target=\"_blank\">@mynkdsi1011</a> , Few Below things you can try .<br>\n1 .  Try Reducing the batch size and epoch.<br>\n2 . If you are using tensorflow try 'Mixed precision' it will help  run faster and use less memory.<br>\n3 . Also as <a href=\"https://www.kaggle.com/vkehfdl1\" target=\"_blank\">@vkehfdl1</a> mentioned split your notebook for EDA+Training+Prediction.<br>\n4 . If you are running this perticular notebook first time try using a smaller model , so that you will get idea what exactly going on.</p>",
      "rawMarkdown": "Hi @mynkdsi1011 , Few Below things you can try .\n1 .  Try Reducing the batch size and epoch.\n2 . If you are using tensorflow try 'Mixed precision' it will help  run faster and use less memory.\n3 . Also as @vkehfdl1 mentioned split your notebook for EDA+Training+Prediction.\n4 . If you are running this perticular notebook first time try using a smaller model , so that you will get idea what exactly going on.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1143795,
      "author_name": "haashaatif",
      "author_url": "",
      "post_date": "01/08/2021 04:24:02",
      "content": "<p>If its a timeout issue, then probably your notebook is running for more than 9 hours at a time which is not allowed as the session limit is 9 hours. If you want to do that, kindly divide it into two parts. 1st is Training and 2nd is Inferring. Do the 1st part in multiple sessions and upload the trained model in some private dataset. Once you get that done, you can use that pretrained model from there in your inference notebook which won't timeout. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1144153,
      "author_name": "vkehfdl1",
      "author_url": "",
      "post_date": "01/08/2021 09:11:50",
      "content": "<p>You can divide your notebook two parts. First, preprocessing and modelling. After you save your model, you can build submission notebook and use model that you saved. You can build your private dataset that save your own models. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1144481,
      "author_name": "ssarkar445",
      "author_url": "",
      "post_date": "01/08/2021 13:36:16",
      "content": "<p>Hi <a href=\"https://www.kaggle.com/mynkdsi1011\" target=\"_blank\">@mynkdsi1011</a> , Few Below things you can try .<br>\n1 .  Try Reducing the batch size and epoch.<br>\n2 . If you are using tensorflow try 'Mixed precision' it will help  run faster and use less memory.<br>\n3 . Also as <a href=\"https://www.kaggle.com/vkehfdl1\" target=\"_blank\">@vkehfdl1</a> mentioned split your notebook for EDA+Training+Prediction.<br>\n4 . If you are running this perticular notebook first time try using a smaller model , so that you will get idea what exactly going on.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1143791": "I am not able to submit my work. I have tried multiple times but notebook is running into timeout issue. Can someone help me here? Thanks in advance. \n\nI am attaching the screen shot.",
    "1143795": "If its a timeout issue, then probably your notebook is running for more than 9 hours at a time which is not allowed as the session limit is 9 hours. If you want to do that, kindly divide it into two parts. 1st is Training and 2nd is Inferring. Do the 1st part in multiple sessions and upload the trained model in some private dataset. Once you get that done, you can use that pretrained model from there in your inference notebook which won't timeout.",
    "1144153": "You can divide your notebook two parts. First, preprocessing and modelling. After you save your model, you can build submission notebook and use model that you saved. You can build your private dataset that save your own models.",
    "1144481": "Hi @mynkdsi1011 , Few Below things you can try .\n1 .  Try Reducing the batch size and epoch.\n2 . If you are using tensorflow try 'Mixed precision' it will help  run faster and use less memory.\n3 . Also as @vkehfdl1 mentioned split your notebook for EDA+Training+Prediction.\n4 . If you are running this perticular notebook first time try using a smaller model , so that you will get idea what exactly going on."
  },
  "source": "meta"
}