{
  "id": 198788,
  "title": "How much time does it take for the submission to show result in a very simple model.",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/198788",
  "author_name": "",
  "post_date": "2020-11-23T03:48:20.619482600Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>The speed is around 25 images/second But the notebook have been running for 3 hours. I dont think it should take that long. Is it normal for it to run that much time?</p>",
  "messages": [
    {
      "id": "1087742",
      "postDate": "11/23/2020 03:48:20",
      "content": "<p>The speed is around 25 images/second But the notebook have been running for 3 hours. I dont think it should take that long. Is it normal for it to run that much time?</p>",
      "rawMarkdown": "The speed is around 25 images/second But the notebook have been running for 3 hours. I dont think it should take that long. Is it normal for it to run that much time?",
      "votes": null
    },
    {
      "id": "1088278",
      "postDate": "11/23/2020 13:45:38",
      "content": "<p>I have exactly the same problem. I've been waiting for the results for a good hour already, after having waited 3 hours to see the commits finish and I wonder if it's a normal behavior.</p>",
      "rawMarkdown": "I have exactly the same problem. I've been waiting for the results for a good hour already, after having waited 3 hours to see the commits finish and I wonder if it's a normal behavior.",
      "votes": null
    },
    {
      "id": "1088282",
      "postDate": "11/23/2020 13:51:26",
      "content": "<p>It must be a bug in your code or something because it takes me less than 10 min to do inference! Good luck</p>",
      "rawMarkdown": "It must be a bug in your code or something because it takes me less than 10 min to do inference! Good luck",
      "votes": null
    },
    {
      "id": "1088530",
      "postDate": "11/23/2020 17:42:37",
      "content": "<p>To make sure your committed notebook can handle the roughly 15,000 test images, I recommend running inference on the train data (21,397 images).</p>\n<p>The provided \"test\" public data is just one image. So that doesn't really test your notebook.</p>\n<p>Everywhere you have \"test\", replace with a variable \"train_or_test\".</p>\n<p>Set:</p>\n<p>train_or_test = 'train'</p>\n<p>for testing.</p>\n<p>Replace sample_submission.csv with train.csv</p>\n<p>Adjust directory names to include your variable, and not be hard-wired to \"test\".</p>\n<p>This will test processing speed, memory utilization, etc.</p>\n<p>Make sure you have internet and TPU turned off. You probably want GPU turned on for inference.</p>\n<p>Adjust your batch size for the maximum you can handle without crashing.</p>\n<p>-Rich</p>",
      "rawMarkdown": "To make sure your committed notebook can handle the roughly 15,000 test images, I recommend running inference on the train data (21,397 images).\n\nThe provided \"test\" public data is just one image. So that doesn't really test your notebook.\n\nEverywhere you have \"test\", replace with a variable \"train_or_test\".\n\nSet:\n\ntrain_or_test = 'train'\n\nfor testing.\n\nReplace sample_submission.csv with train.csv\n\nAdjust directory names to include your variable, and not be hard-wired to \"test\".\n\nThis will test processing speed, memory utilization, etc.\n\nMake sure you have internet and TPU turned off. You probably want GPU turned on for inference.\n\nAdjust your batch size for the maximum you can handle without crashing.\n\n-Rich",
      "votes": null
    },
    {
      "id": "1110525",
      "postDate": "12/12/2020 20:35:02",
      "content": "<p>wondering if you were able to fix your issue. i have the same issue. </p>\n<ol>\n<li>without gpu , the notebook is still running after 2 hours</li>\n<li>as suggested i tried switching on test to train and turned on gpu , that took like a hour for inference to run thru 26k records </li>\n</ol>",
      "rawMarkdown": "wondering if you were able to fix your issue. i have the same issue. \n1. without gpu , the notebook is still running after 2 hours\n2. as suggested i tried switching on test to train and turned on gpu , that took like a hour for inference to run thru 26k records",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1088278,
      "author_name": "danielarmel",
      "author_url": "",
      "post_date": "11/23/2020 13:45:38",
      "content": "<p>I have exactly the same problem. I've been waiting for the results for a good hour already, after having waited 3 hours to see the commits finish and I wonder if it's a normal behavior.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1110525,
          "author_name": "venkat555",
          "author_url": "",
          "post_date": "12/12/2020 20:35:02",
          "content": "<p>wondering if you were able to fix your issue. i have the same issue. </p>\n<ol>\n<li>without gpu , the notebook is still running after 2 hours</li>\n<li>as suggested i tried switching on test to train and turned on gpu , that took like a hour for inference to run thru 26k records </li>\n</ol>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1088282,
      "author_name": "yannmajewski",
      "author_url": "",
      "post_date": "11/23/2020 13:51:26",
      "content": "<p>It must be a bug in your code or something because it takes me less than 10 min to do inference! Good luck</p>",
      "votes": null,
      "replies": [
        {
          "id": 1088530,
          "author_name": "richardepstein",
          "author_url": "",
          "post_date": "11/23/2020 17:42:37",
          "content": "<p>To make sure your committed notebook can handle the roughly 15,000 test images, I recommend running inference on the train data (21,397 images).</p>\n<p>The provided \"test\" public data is just one image. So that doesn't really test your notebook.</p>\n<p>Everywhere you have \"test\", replace with a variable \"train_or_test\".</p>\n<p>Set:</p>\n<p>train_or_test = 'train'</p>\n<p>for testing.</p>\n<p>Replace sample_submission.csv with train.csv</p>\n<p>Adjust directory names to include your variable, and not be hard-wired to \"test\".</p>\n<p>This will test processing speed, memory utilization, etc.</p>\n<p>Make sure you have internet and TPU turned off. You probably want GPU turned on for inference.</p>\n<p>Adjust your batch size for the maximum you can handle without crashing.</p>\n<p>-Rich</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1087742": "The speed is around 25 images/second But the notebook have been running for 3 hours. I dont think it should take that long. Is it normal for it to run that much time?",
    "1088278": "I have exactly the same problem. I've been waiting for the results for a good hour already, after having waited 3 hours to see the commits finish and I wonder if it's a normal behavior.",
    "1088282": "It must be a bug in your code or something because it takes me less than 10 min to do inference! Good luck",
    "1088530": "To make sure your committed notebook can handle the roughly 15,000 test images, I recommend running inference on the train data (21,397 images).\n\nThe provided \"test\" public data is just one image. So that doesn't really test your notebook.\n\nEverywhere you have \"test\", replace with a variable \"train_or_test\".\n\nSet:\n\ntrain_or_test = 'train'\n\nfor testing.\n\nReplace sample_submission.csv with train.csv\n\nAdjust directory names to include your variable, and not be hard-wired to \"test\".\n\nThis will test processing speed, memory utilization, etc.\n\nMake sure you have internet and TPU turned off. You probably want GPU turned on for inference.\n\nAdjust your batch size for the maximum you can handle without crashing.\n\n-Rich",
    "1110525": "wondering if you were able to fix your issue. i have the same issue. \n1. without gpu , the notebook is still running after 2 hours\n2. as suggested i tried switching on test to train and turned on gpu , that took like a hour for inference to run thru 26k records"
  },
  "source": "meta"
}