{
  "id": 216315,
  "title": "Why is it stipulated that the pre training model can be used, but not the Internet?",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/216315",
  "author_name": "",
  "post_date": "2021-02-02T11:56:06.671522900Z",
  "votes": 1,
  "comment_count": 15,
  "views": 0,
  "content": "<p>！！！！Who knows?</p>",
  "messages": [
    {
      "id": "1182287",
      "postDate": "02/02/2021 11:56:06",
      "content": "<p>！！！！Who knows?</p>",
      "rawMarkdown": "！！！！Who knows?",
      "votes": null
    },
    {
      "id": "1182325",
      "postDate": "02/02/2021 12:14:55",
      "content": "<p>Probably for security reasons.</p>",
      "rawMarkdown": "Probably for security reasons.",
      "votes": null
    },
    {
      "id": "1182329",
      "postDate": "02/02/2021 12:17:35",
      "content": "<p>However, the pretraining model cannot be used without Internet.  Is the title rule wrong, or do I misunderstand it?</p>",
      "rawMarkdown": "However, the pretraining model cannot be used without Internet.  Is the title rule wrong, or do I misunderstand it?",
      "votes": null
    },
    {
      "id": "1182339",
      "postDate": "02/02/2021 12:20:31",
      "content": "<p>Because it's not a big deal to write a script to \"steal\" the hidden set having internet turned on in the notebook.</p>",
      "rawMarkdown": "Because it's not a big deal to write a script to \"steal\" the hidden set having internet turned on in the notebook.",
      "votes": null
    },
    {
      "id": "1182342",
      "postDate": "02/02/2021 12:22:17",
      "content": "<p>Can I use the pre training model? How to import it?</p>",
      "rawMarkdown": "Can I use the pre training model? How to import it?",
      "votes": null
    },
    {
      "id": "1182344",
      "postDate": "02/02/2021 12:22:50",
      "content": "<p>If you use <code>efficientnet</code> or other models which does not come preinstalled with Keras/Pytorch, you can add a dataset with a source code to your notebook and continue without an internet connection.</p>",
      "rawMarkdown": "If you use `efficientnet` or other models which does not come preinstalled with Keras/Pytorch, you can add a dataset with a source code to your notebook and continue without an internet connection.",
      "votes": null
    },
    {
      "id": "1182349",
      "postDate": "02/02/2021 12:24:43",
      "content": "<p>Sure. I've accidentally replied below</p>",
      "rawMarkdown": "Sure. I've accidentally replied below",
      "votes": null
    },
    {
      "id": "1182396",
      "postDate": "02/02/2021 12:50:02",
      "content": "<p>Thank you for your reply. dataset with a source code is it?   Do I download the pre training model first and then upload it to the code?</p>",
      "rawMarkdown": "Thank you for your reply. dataset with a source code is it?   Do I download the pre training model first and then upload it to the code?",
      "votes": null
    },
    {
      "id": "1182414",
      "postDate": "02/02/2021 12:56:00",
      "content": "<p>You can enable internet and use pretrained weight for training</p>\n<p>You don't need the pretrained weight (nor internet) for inference. Given you've already trained weight for your model. </p>\n<p>I don't know why these questions are asked multiple times a day with new topic while a simple search give the right answer. </p>",
      "rawMarkdown": "You can enable internet and use pretrained weight for training\n\nYou don't need the pretrained weight (nor internet) for inference. Given you've already trained weight for your model. \n\nI don't know why these questions are asked multiple times a day with new topic while a simple search give the right answer.",
      "votes": null
    },
    {
      "id": "1182438",
      "postDate": "02/02/2021 13:05:10",
      "content": "<p>Generally speaking, firstly you train your model in your train notebook (where you <code>pip install</code> all the dependencies which aren't preinstalled on Kaggle kernels), then you save your model to a private dataset and add it (as well as the datasets of source code of all missing dependencies) to your inference notebook without an internet connection.</p>",
      "rawMarkdown": "Generally speaking, firstly you train your model in your train notebook (where you `pip install` all the dependencies which aren't preinstalled on Kaggle kernels), then you save your model to a private dataset and add it (as well as the datasets of source code of all missing dependencies) to your inference notebook without an internet connection.",
      "votes": null
    },
    {
      "id": "1182483",
      "postDate": "02/02/2021 13:26:59",
      "content": "<p>When I submit：Cannot submit  Your Notebook cannot use internet access in this competition. Please disable internet in the Notebook editor and save a new version<br>\nWhen I close the Internet and save it, the parameters of the last training are gone</p>",
      "rawMarkdown": "When I submit：Cannot submit  Your Notebook cannot use internet access in this competition. Please disable internet in the Notebook editor and save a new version\nWhen I close the Internet and save it, the parameters of the last training are gone",
      "votes": null
    },
    {
      "id": "1182485",
      "postDate": "02/02/2021 13:27:41",
      "content": "<p>Thank you for your reply</p>",
      "rawMarkdown": "Thank you for your reply",
      "votes": null
    },
    {
      "id": "1182521",
      "postDate": "02/02/2021 13:39:02",
      "content": "<p>You need to train on one notebook (with internet on) and do inference  one another ( with internet off and trained weight uploaded as dataset)</p>\n<p>Or if you train outside of kaggle as I do, just upload the weights as dataset and do inference without internet </p>",
      "rawMarkdown": "You need to train on one notebook (with internet on) and do inference  one another ( with internet off and trained weight uploaded as dataset)\n\nOr if you train outside of kaggle as I do, just upload the weights as dataset and do inference without internet",
      "votes": null
    },
    {
      "id": "1182525",
      "postDate": "02/02/2021 13:40:47",
      "content": "<p>I totally understand.Thank you for your reply. I'll try</p>",
      "rawMarkdown": "I totally understand.Thank you for your reply. I'll try",
      "votes": null
    },
    {
      "id": "1182527",
      "postDate": "02/02/2021 13:41:34",
      "content": "<p>I totally understand.Thank you for your reply. I'll try.Thanks again</p>",
      "rawMarkdown": "I totally understand.Thank you for your reply. I'll try.Thanks again",
      "votes": null
    },
    {
      "id": "1187293",
      "postDate": "02/05/2021 10:52:10",
      "content": "<p>check my  notebook to learn how to use  pre-trained models without internet<br>\n<a href=\"https://www.kaggle.com/iraqai/keras-efficientnet7-aug-tpu-training-88\" target=\"_blank\">https://www.kaggle.com/iraqai/keras-efficientnet7-aug-tpu-training-88</a></p>",
      "rawMarkdown": "check my  notebook to learn how to use  pre-trained models without internet\nhttps://www.kaggle.com/iraqai/keras-efficientnet7-aug-tpu-training-88",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1182325,
      "author_name": "killimi",
      "author_url": "",
      "post_date": "02/02/2021 12:14:55",
      "content": "<p>Probably for security reasons.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1182329,
          "author_name": "henini",
          "author_url": "",
          "post_date": "02/02/2021 12:17:35",
          "content": "<p>However, the pretraining model cannot be used without Internet.  Is the title rule wrong, or do I misunderstand it?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182344,
          "author_name": "nickuzmenkov",
          "author_url": "",
          "post_date": "02/02/2021 12:22:50",
          "content": "<p>If you use <code>efficientnet</code> or other models which does not come preinstalled with Keras/Pytorch, you can add a dataset with a source code to your notebook and continue without an internet connection.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1182339,
      "author_name": "nickuzmenkov",
      "author_url": "",
      "post_date": "02/02/2021 12:20:31",
      "content": "<p>Because it's not a big deal to write a script to \"steal\" the hidden set having internet turned on in the notebook.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1182342,
          "author_name": "henini",
          "author_url": "",
          "post_date": "02/02/2021 12:22:17",
          "content": "<p>Can I use the pre training model? How to import it?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182349,
          "author_name": "nickuzmenkov",
          "author_url": "",
          "post_date": "02/02/2021 12:24:43",
          "content": "<p>Sure. I've accidentally replied below</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182396,
          "author_name": "henini",
          "author_url": "",
          "post_date": "02/02/2021 12:50:02",
          "content": "<p>Thank you for your reply. dataset with a source code is it?   Do I download the pre training model first and then upload it to the code?</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182438,
          "author_name": "nickuzmenkov",
          "author_url": "",
          "post_date": "02/02/2021 13:05:10",
          "content": "<p>Generally speaking, firstly you train your model in your train notebook (where you <code>pip install</code> all the dependencies which aren't preinstalled on Kaggle kernels), then you save your model to a private dataset and add it (as well as the datasets of source code of all missing dependencies) to your inference notebook without an internet connection.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182525,
          "author_name": "henini",
          "author_url": "",
          "post_date": "02/02/2021 13:40:47",
          "content": "<p>I totally understand.Thank you for your reply. I'll try</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1182414,
      "author_name": "serigne",
      "author_url": "",
      "post_date": "02/02/2021 12:56:00",
      "content": "<p>You can enable internet and use pretrained weight for training</p>\n<p>You don't need the pretrained weight (nor internet) for inference. Given you've already trained weight for your model. </p>\n<p>I don't know why these questions are asked multiple times a day with new topic while a simple search give the right answer. </p>",
      "votes": null,
      "replies": [
        {
          "id": 1182483,
          "author_name": "henini",
          "author_url": "",
          "post_date": "02/02/2021 13:26:59",
          "content": "<p>When I submit：Cannot submit  Your Notebook cannot use internet access in this competition. Please disable internet in the Notebook editor and save a new version<br>\nWhen I close the Internet and save it, the parameters of the last training are gone</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182485,
          "author_name": "henini",
          "author_url": "",
          "post_date": "02/02/2021 13:27:41",
          "content": "<p>Thank you for your reply</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182521,
          "author_name": "serigne",
          "author_url": "",
          "post_date": "02/02/2021 13:39:02",
          "content": "<p>You need to train on one notebook (with internet on) and do inference  one another ( with internet off and trained weight uploaded as dataset)</p>\n<p>Or if you train outside of kaggle as I do, just upload the weights as dataset and do inference without internet </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1182527,
          "author_name": "henini",
          "author_url": "",
          "post_date": "02/02/2021 13:41:34",
          "content": "<p>I totally understand.Thank you for your reply. I'll try.Thanks again</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1187293,
      "author_name": "iraqai",
      "author_url": "",
      "post_date": "02/05/2021 10:52:10",
      "content": "<p>check my  notebook to learn how to use  pre-trained models without internet<br>\n<a href=\"https://www.kaggle.com/iraqai/keras-efficientnet7-aug-tpu-training-88\" target=\"_blank\">https://www.kaggle.com/iraqai/keras-efficientnet7-aug-tpu-training-88</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1182287": "！！！！Who knows?",
    "1182325": "Probably for security reasons.",
    "1182329": "However, the pretraining model cannot be used without Internet.  Is the title rule wrong, or do I misunderstand it?",
    "1182339": "Because it's not a big deal to write a script to \"steal\" the hidden set having internet turned on in the notebook.",
    "1182342": "Can I use the pre training model? How to import it?",
    "1182344": "If you use `efficientnet` or other models which does not come preinstalled with Keras/Pytorch, you can add a dataset with a source code to your notebook and continue without an internet connection.",
    "1182349": "Sure. I've accidentally replied below",
    "1182396": "Thank you for your reply. dataset with a source code is it?   Do I download the pre training model first and then upload it to the code?",
    "1182414": "You can enable internet and use pretrained weight for training\n\nYou don't need the pretrained weight (nor internet) for inference. Given you've already trained weight for your model. \n\nI don't know why these questions are asked multiple times a day with new topic while a simple search give the right answer.",
    "1182438": "Generally speaking, firstly you train your model in your train notebook (where you `pip install` all the dependencies which aren't preinstalled on Kaggle kernels), then you save your model to a private dataset and add it (as well as the datasets of source code of all missing dependencies) to your inference notebook without an internet connection.",
    "1182483": "When I submit：Cannot submit  Your Notebook cannot use internet access in this competition. Please disable internet in the Notebook editor and save a new version\nWhen I close the Internet and save it, the parameters of the last training are gone",
    "1182485": "Thank you for your reply",
    "1182521": "You need to train on one notebook (with internet on) and do inference  one another ( with internet off and trained weight uploaded as dataset)\n\nOr if you train outside of kaggle as I do, just upload the weights as dataset and do inference without internet",
    "1182525": "I totally understand.Thank you for your reply. I'll try",
    "1182527": "I totally understand.Thank you for your reply. I'll try.Thanks again",
    "1187293": "check my  notebook to learn how to use  pre-trained models without internet\nhttps://www.kaggle.com/iraqai/keras-efficientnet7-aug-tpu-training-88"
  },
  "source": "meta"
}