{
  "id": 214634,
  "title": "submission taking too much time for showing public score, need help!",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/214634",
  "author_name": "",
  "post_date": "2021-01-27T08:31:05.152983100Z",
  "votes": 3,
  "comment_count": 6,
  "views": 0,
  "content": "<p>I submitted my final submission.csv file but it is showing '' Our workers are spending a little extra time on your submission. Grab a coffee and refresh later to see your score.'' is there any thing wrong or this is usually happens some times.</p>",
  "messages": [
    {
      "id": "1172046",
      "postDate": "01/27/2021 08:31:05",
      "content": "<p>I submitted my final submission.csv file but it is showing '' Our workers are spending a little extra time on your submission. Grab a coffee and refresh later to see your score.'' is there any thing wrong or this is usually happens some times.</p>",
      "rawMarkdown": "I submitted my final submission.csv file but it is showing '' Our workers are spending a little extra time on your submission. Grab a coffee and refresh later to see your score.'' is there any thing wrong or this is usually happens some times.",
      "votes": null
    },
    {
      "id": "1172052",
      "postDate": "01/27/2021 08:35:04",
      "content": "<p>If your submission notebook includes training process, it will takes lots of time. Because whole notebook run again and save outputs. So you can split training notebook and submission notebook. In keras, you can save your model using ModelCheckpoint callbacks, and load your file using tf.keras.models.load_model. </p>",
      "rawMarkdown": "If your submission notebook includes training process, it will takes lots of time. Because whole notebook run again and save outputs. So you can split training notebook and submission notebook. In keras, you can save your model using ModelCheckpoint callbacks, and load your file using tf.keras.models.load_model.",
      "votes": null
    },
    {
      "id": "1172590",
      "postDate": "01/27/2021 12:57:04",
      "content": "<p>Hello <a href=\"https://www.kaggle.com/kamleshshimpi78\" target=\"_blank\">@kamleshshimpi78</a> ,<br>\nHere are some tips for debugging your submission:</p>\n<ul>\n<li>If you are using TTA, try to set tta_steps to 1 .</li>\n<li>if you are using multiple models, try using only single model.</li>\n<li>Also seperate the training and inference process in two different notebooks.</li>\n</ul>\n<p>In short, try to submit using single model and <code>tta_steps = 1</code>.</p>\n<p>With these, you will be able to know, what thing is eating your time.</p>",
      "rawMarkdown": "Hello @kamleshshimpi78 ,\nHere are some tips for debugging your submission:\n\n- If you are using TTA, try to set tta_steps to 1 .\n- if you are using multiple models, try using only single model.\n- Also seperate the training and inference process in two different notebooks.\n\nIn short, try to submit using single model and `tta_steps = 1`.\n\nWith these, you will be able to know, what thing is eating your time.",
      "votes": null
    },
    {
      "id": "1172594",
      "postDate": "01/27/2021 12:58:40",
      "content": "<p>Mine take around 34 min using single renet34 model with no TTA. </p>",
      "rawMarkdown": "Mine take around 34 min using single renet34 model with no TTA.",
      "votes": null
    },
    {
      "id": "1172663",
      "postDate": "01/27/2021 13:44:50",
      "content": "<p>It is pretty normal </p>",
      "rawMarkdown": "It is pretty normal",
      "votes": null
    },
    {
      "id": "1173264",
      "postDate": "01/27/2021 18:55:48",
      "content": "<p>Normal,don’t worry about it </p>",
      "rawMarkdown": "Normal,don’t worry about it",
      "votes": null
    },
    {
      "id": "1174827",
      "postDate": "01/28/2021 18:29:54",
      "content": "<p>Do not worry Pal.<br>\nIt is just running your whole notebook all over again.<br>\nBut this time, automatically your notebook will get around 1.5k images<br>\nin test_images folder.<br>\nand the results will be calculated for these images.<br>\nSo, nothing to worry, just grab a coffee and wait for a little time.</p>",
      "rawMarkdown": "Do not worry Pal.\nIt is just running your whole notebook all over again.\nBut this time, automatically your notebook will get around 1.5k images\nin test_images folder.\nand the results will be calculated for these images.\nSo, nothing to worry, just grab a coffee and wait for a little time.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1172052,
      "author_name": "vkehfdl1",
      "author_url": "",
      "post_date": "01/27/2021 08:35:04",
      "content": "<p>If your submission notebook includes training process, it will takes lots of time. Because whole notebook run again and save outputs. So you can split training notebook and submission notebook. In keras, you can save your model using ModelCheckpoint callbacks, and load your file using tf.keras.models.load_model. </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1172590,
      "author_name": "joshi98kishan",
      "author_url": "",
      "post_date": "01/27/2021 12:57:04",
      "content": "<p>Hello <a href=\"https://www.kaggle.com/kamleshshimpi78\" target=\"_blank\">@kamleshshimpi78</a> ,<br>\nHere are some tips for debugging your submission:</p>\n<ul>\n<li>If you are using TTA, try to set tta_steps to 1 .</li>\n<li>if you are using multiple models, try using only single model.</li>\n<li>Also seperate the training and inference process in two different notebooks.</li>\n</ul>\n<p>In short, try to submit using single model and <code>tta_steps = 1</code>.</p>\n<p>With these, you will be able to know, what thing is eating your time.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1172594,
          "author_name": "joshi98kishan",
          "author_url": "",
          "post_date": "01/27/2021 12:58:40",
          "content": "<p>Mine take around 34 min using single renet34 model with no TTA. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1172663,
      "author_name": "mithilsalunkhe",
      "author_url": "",
      "post_date": "01/27/2021 13:44:50",
      "content": "<p>It is pretty normal </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1173264,
      "author_name": "darknesszx",
      "author_url": "",
      "post_date": "01/27/2021 18:55:48",
      "content": "<p>Normal,don’t worry about it </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1174827,
      "author_name": "ayannareda",
      "author_url": "",
      "post_date": "01/28/2021 18:29:54",
      "content": "<p>Do not worry Pal.<br>\nIt is just running your whole notebook all over again.<br>\nBut this time, automatically your notebook will get around 1.5k images<br>\nin test_images folder.<br>\nand the results will be calculated for these images.<br>\nSo, nothing to worry, just grab a coffee and wait for a little time.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1172046": "I submitted my final submission.csv file but it is showing '' Our workers are spending a little extra time on your submission. Grab a coffee and refresh later to see your score.'' is there any thing wrong or this is usually happens some times.",
    "1172052": "If your submission notebook includes training process, it will takes lots of time. Because whole notebook run again and save outputs. So you can split training notebook and submission notebook. In keras, you can save your model using ModelCheckpoint callbacks, and load your file using tf.keras.models.load_model.",
    "1172590": "Hello @kamleshshimpi78 ,\nHere are some tips for debugging your submission:\n\n- If you are using TTA, try to set tta_steps to 1 .\n- if you are using multiple models, try using only single model.\n- Also seperate the training and inference process in two different notebooks.\n\nIn short, try to submit using single model and `tta_steps = 1`.\n\nWith these, you will be able to know, what thing is eating your time.",
    "1172594": "Mine take around 34 min using single renet34 model with no TTA.",
    "1172663": "It is pretty normal",
    "1173264": "Normal,don’t worry about it",
    "1174827": "Do not worry Pal.\nIt is just running your whole notebook all over again.\nBut this time, automatically your notebook will get around 1.5k images\nin test_images folder.\nand the results will be calculated for these images.\nSo, nothing to worry, just grab a coffee and wait for a little time."
  },
  "source": "meta"
}