{
  "id": 267417,
  "title": "How to use my model on the secret Dataset?",
  "url": "/competitions/landmark-recognition-2021/discussion/267417",
  "author_name": "",
  "post_date": "2021-08-23T06:43:56.753594700Z",
  "votes": 4,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I've saved my model as .h5 format and also saved a numpy array which I need for the model as .h5 format also.<br>\nThen I load them in my evaluation code which I try to use for my submission.<br>\nI always get the error of <code>Notebook Threw Exception</code> and I don't seem to understand where the problem might be.<br>\nWhile using my code on the visible test files it seems to work fine, but when the model is trying to predict the hidden test data the problem occurs.</p>\n<p>Any suggestions might be helpful,</p>\n<p>Thank you</p>",
  "messages": [
    {
      "id": "1486683",
      "postDate": "08/23/2021 06:43:56",
      "content": "<p>I've saved my model as .h5 format and also saved a numpy array which I need for the model as .h5 format also.<br>\nThen I load them in my evaluation code which I try to use for my submission.<br>\nI always get the error of <code>Notebook Threw Exception</code> and I don't seem to understand where the problem might be.<br>\nWhile using my code on the visible test files it seems to work fine, but when the model is trying to predict the hidden test data the problem occurs.</p>\n<p>Any suggestions might be helpful,</p>\n<p>Thank you</p>",
      "rawMarkdown": "I've saved my model as .h5 format and also saved a numpy array which I need for the model as .h5 format also.\nThen I load them in my evaluation code which I try to use for my submission.\nI always get the error of ```Notebook Threw Exception``` and I don't seem to understand where the problem might be.\nWhile using my code on the visible test files it seems to work fine, but when the model is trying to predict the hidden test data the problem occurs.\n\nAny suggestions might be helpful,\n\nThank you",
      "votes": null
    },
    {
      "id": "1488563",
      "postDate": "08/24/2021 11:26:16",
      "content": "<p>I load a pretrained keras model from a public dataset and tune it in the notebook. Everything runs fine and I'm able to save the versions. But when I commit them, I get the same error. Check out this discussion: <a href=\"https://www.kaggle.com/c/landmark-recognition-2021/discussion/264799\" target=\"_blank\">https://www.kaggle.com/c/landmark-recognition-2021/discussion/264799</a><br>\nI was unable to decide what to do after reading this. How are we supposed to train the model without reading \"train.csv\" or any other files from the \"train\" folder?</p>",
      "rawMarkdown": "I load a pretrained keras model from a public dataset and tune it in the notebook. Everything runs fine and I'm able to save the versions. But when I commit them, I get the same error. Check out this discussion: https://www.kaggle.com/c/landmark-recognition-2021/discussion/264799\nI was unable to decide what to do after reading this. How are we supposed to train the model without reading \"train.csv\" or any other files from the \"train\" folder?",
      "votes": null
    },
    {
      "id": "1490621",
      "postDate": "08/25/2021 18:21:16",
      "content": "<p>Exactly,<br>\nIf someone managed to understand how to deal with it we will be happy for advice</p>",
      "rawMarkdown": "Exactly,\nIf someone managed to understand how to deal with it we will be happy for advice",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1488563,
      "author_name": "ekaterinadranitsyna",
      "author_url": "",
      "post_date": "08/24/2021 11:26:16",
      "content": "<p>I load a pretrained keras model from a public dataset and tune it in the notebook. Everything runs fine and I'm able to save the versions. But when I commit them, I get the same error. Check out this discussion: <a href=\"https://www.kaggle.com/c/landmark-recognition-2021/discussion/264799\" target=\"_blank\">https://www.kaggle.com/c/landmark-recognition-2021/discussion/264799</a><br>\nI was unable to decide what to do after reading this. How are we supposed to train the model without reading \"train.csv\" or any other files from the \"train\" folder?</p>",
      "votes": null,
      "replies": [
        {
          "id": 1490621,
          "author_name": "nizado",
          "author_url": "",
          "post_date": "08/25/2021 18:21:16",
          "content": "<p>Exactly,<br>\nIf someone managed to understand how to deal with it we will be happy for advice</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1486683": "I've saved my model as .h5 format and also saved a numpy array which I need for the model as .h5 format also.\nThen I load them in my evaluation code which I try to use for my submission.\nI always get the error of ```Notebook Threw Exception``` and I don't seem to understand where the problem might be.\nWhile using my code on the visible test files it seems to work fine, but when the model is trying to predict the hidden test data the problem occurs.\n\nAny suggestions might be helpful,\n\nThank you",
    "1488563": "I load a pretrained keras model from a public dataset and tune it in the notebook. Everything runs fine and I'm able to save the versions. But when I commit them, I get the same error. Check out this discussion: https://www.kaggle.com/c/landmark-recognition-2021/discussion/264799\nI was unable to decide what to do after reading this. How are we supposed to train the model without reading \"train.csv\" or any other files from the \"train\" folder?",
    "1490621": "Exactly,\nIf someone managed to understand how to deal with it we will be happy for advice"
  },
  "source": "meta"
}