{
  "id": 131891,
  "title": "example of a notebook that works a submission",
  "url": "/competitions/bengaliai-cv19/discussion/131891",
  "author_name": "",
  "post_date": "2020-02-22T12:30:02.378960700Z",
  "votes": null,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi,</p>\n\n<p>Can I have a example of a notebook which is ok to make a submission ?</p>\n\n<p>Thanks</p>",
  "messages": [
    {
      "id": "753589",
      "postDate": "02/22/2020 12:30:02",
      "content": "<p>Hi,</p>\n\n<p>Can I have a example of a notebook which is ok to make a submission ?</p>\n\n<p>Thanks</p>",
      "rawMarkdown": "Hi,\n\nCan I have a example of a notebook which is ok to make a submission ?\n\nThanks",
      "votes": null
    },
    {
      "id": "753642",
      "postDate": "02/22/2020 13:54:57",
      "content": "<p><a href=\"https://www.kaggle.com/h030162/version1-0-9696\">This</a> is a great kernel for making submissions.\nThe main thing to notice is that the user has made use of pre trained weights, meaning that the model does not train in this notebook, rather creates the model architecture, loads the weights and generates the predictions. This way you can keep the runtime down during submission.\nThe user had made use of a different (private) notebook for training purposes. Once a good accuracy was achieved, he saved the model weights into s file which he then loads into the submission notebook for validation.</p>",
      "rawMarkdown": "[This](https://www.kaggle.com/h030162/version1-0-9696) is a great kernel for making submissions.\nThe main thing to notice is that the user has made use of pre trained weights, meaning that the model does not train in this notebook, rather creates the model architecture, loads the weights and generates the predictions. This way you can keep the runtime down during submission.\nThe user had made use of a different (private) notebook for training purposes. Once a good accuracy was achieved, he saved the model weights into s file which he then loads into the submission notebook for validation.",
      "votes": null
    },
    {
      "id": "754440",
      "postDate": "02/23/2020 14:53:36",
      "content": "<p>If I download this kernel and I create a submission with it, I will have a score on my own account ? Sorry, I am quite new for this way of submitting.</p>",
      "rawMarkdown": "If I download this kernel and I create a submission with it, I will have a score on my own account ? Sorry, I am quite new for this way of submitting.",
      "votes": null
    },
    {
      "id": "754444",
      "postDate": "02/23/2020 14:58:03",
      "content": "<p>Or perhaps I just have to upload my weights and make predictions with ?</p>",
      "rawMarkdown": "Or perhaps I just have to upload my weights and make predictions with ?",
      "votes": null
    },
    {
      "id": "754496",
      "postDate": "02/23/2020 16:25:40",
      "content": "<p>If you submit the same code without changing a line, you will get the same score (in your account).\n(I assume you have access to the weights as well)</p>",
      "rawMarkdown": "If you submit the same code without changing a line, you will get the same score (in your account).\n(I assume you have access to the weights as well)",
      "votes": null
    },
    {
      "id": "754498",
      "postDate": "02/23/2020 16:28:21",
      "content": "<p>Exactly. If you trained your model offline, you can upload the weights to a private dataset. Because this is a kernel-only competition, you have to do the prediction from a kernel.</p>",
      "rawMarkdown": "Exactly. If you trained your model offline, you can upload the weights to a private dataset. Because this is a kernel-only competition, you have to do the prediction from a kernel.",
      "votes": null
    },
    {
      "id": "754588",
      "postDate": "02/23/2020 19:38:07",
      "content": "<p>..  hmmm  ... Can you give me a python code witch only loads some keras weights and builds a csv file with results in ?</p>",
      "rawMarkdown": "..  hmmm  ... Can you give me a python code witch only loads some keras weights and builds a csv file with results in ?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 753642,
      "author_name": "skillsmuggler",
      "author_url": "",
      "post_date": "02/22/2020 13:54:57",
      "content": "<p><a href=\"https://www.kaggle.com/h030162/version1-0-9696\">This</a> is a great kernel for making submissions.\nThe main thing to notice is that the user has made use of pre trained weights, meaning that the model does not train in this notebook, rather creates the model architecture, loads the weights and generates the predictions. This way you can keep the runtime down during submission.\nThe user had made use of a different (private) notebook for training purposes. Once a good accuracy was achieved, he saved the model weights into s file which he then loads into the submission notebook for validation.</p>",
      "votes": null,
      "replies": [
        {
          "id": 754440,
          "author_name": "johnnyhalliday",
          "author_url": "",
          "post_date": "02/23/2020 14:53:36",
          "content": "<p>If I download this kernel and I create a submission with it, I will have a score on my own account ? Sorry, I am quite new for this way of submitting.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 754496,
          "author_name": "pestipeti",
          "author_url": "",
          "post_date": "02/23/2020 16:25:40",
          "content": "<p>If you submit the same code without changing a line, you will get the same score (in your account).\n(I assume you have access to the weights as well)</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 754444,
      "author_name": "johnnyhalliday",
      "author_url": "",
      "post_date": "02/23/2020 14:58:03",
      "content": "<p>Or perhaps I just have to upload my weights and make predictions with ?</p>",
      "votes": null,
      "replies": [
        {
          "id": 754498,
          "author_name": "pestipeti",
          "author_url": "",
          "post_date": "02/23/2020 16:28:21",
          "content": "<p>Exactly. If you trained your model offline, you can upload the weights to a private dataset. Because this is a kernel-only competition, you have to do the prediction from a kernel.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 754588,
          "author_name": "johnnyhalliday",
          "author_url": "",
          "post_date": "02/23/2020 19:38:07",
          "content": "<p>..  hmmm  ... Can you give me a python code witch only loads some keras weights and builds a csv file with results in ?</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "753589": "Hi,\n\nCan I have a example of a notebook which is ok to make a submission ?\n\nThanks",
    "753642": "[This](https://www.kaggle.com/h030162/version1-0-9696) is a great kernel for making submissions.\nThe main thing to notice is that the user has made use of pre trained weights, meaning that the model does not train in this notebook, rather creates the model architecture, loads the weights and generates the predictions. This way you can keep the runtime down during submission.\nThe user had made use of a different (private) notebook for training purposes. Once a good accuracy was achieved, he saved the model weights into s file which he then loads into the submission notebook for validation.",
    "754440": "If I download this kernel and I create a submission with it, I will have a score on my own account ? Sorry, I am quite new for this way of submitting.",
    "754444": "Or perhaps I just have to upload my weights and make predictions with ?",
    "754496": "If you submit the same code without changing a line, you will get the same score (in your account).\n(I assume you have access to the weights as well)",
    "754498": "Exactly. If you trained your model offline, you can upload the weights to a private dataset. Because this is a kernel-only competition, you have to do the prediction from a kernel.",
    "754588": "..  hmmm  ... Can you give me a python code witch only loads some keras weights and builds a csv file with results in ?"
  },
  "source": "meta"
}