{
  "id": 122913,
  "title": "Private notebook rule",
  "url": "/competitions/bengaliai-cv19/discussion/122913",
  "author_name": "",
  "post_date": "2019-12-23T16:14:47.274847500Z",
  "votes": 1,
  "comment_count": 6,
  "views": 0,
  "content": "<p>Hi competition hosts,\n<a href=\"https://www.kaggle.com/c/bengaliai-cv19/overview/notebooks-requirements\">https://www.kaggle.com/c/bengaliai-cv19/overview/notebooks-requirements</a>\nHere says \"you are encouraged to train your model offline and use your Notebook for inference\".\nDoes it means that I can train my model offline and I can keep the model weight in private even though the model is trained on external data, am I correct? (I know using external data should disclose in the thread)</p>",
  "messages": [
    {
      "id": "701570",
      "postDate": "12/23/2019 16:14:47",
      "content": "<p>Hi competition hosts,\n<a href=\"https://www.kaggle.com/c/bengaliai-cv19/overview/notebooks-requirements\">https://www.kaggle.com/c/bengaliai-cv19/overview/notebooks-requirements</a>\nHere says \"you are encouraged to train your model offline and use your Notebook for inference\".\nDoes it means that I can train my model offline and I can keep the model weight in private even though the model is trained on external data, am I correct? (I know using external data should disclose in the thread)</p>",
      "rawMarkdown": "Hi competition hosts,\nhttps://www.kaggle.com/c/bengaliai-cv19/overview/notebooks-requirements\nHere says \"you are encouraged to train your model offline and use your Notebook for inference\".\nDoes it means that I can train my model offline and I can keep the model weight in private even though the model is trained on external data, am I correct? (I know using external data should disclose in the thread)",
      "votes": null
    },
    {
      "id": "701583",
      "postDate": "12/23/2019 16:26:03",
      "content": "<p>Yes </p>",
      "rawMarkdown": "Yes",
      "votes": null
    },
    {
      "id": "701594",
      "postDate": "12/23/2019 16:38:01",
      "content": "<p>Same question here:\nif I want to access my pretrained model weights from my notebook (if they don't fit into a notebook), do I need to give public access to the dataset containing the weights? </p>",
      "rawMarkdown": "Same question here:\nif I want to access my pretrained model weights from my notebook (if they don't fit into a notebook), do I need to give public access to the dataset containing the weights?",
      "votes": null
    },
    {
      "id": "701605",
      "postDate": "12/23/2019 16:48:21",
      "content": "<p>No you don’t. You have Access to 20gb of private storage.</p>",
      "rawMarkdown": "No you don’t. You have Access to 20gb of private storage.",
      "votes": null
    },
    {
      "id": "701633",
      "postDate": "12/23/2019 17:28:38",
      "content": "<p>Correct - you can train offline and upload your model weights to a private notebook for inference. </p>",
      "rawMarkdown": "Correct - you can train offline and upload your model weights to a private notebook for inference.",
      "votes": null
    },
    {
      "id": "701807",
      "postDate": "12/24/2019 00:11:10",
      "content": "<p>Thank you!</p>",
      "rawMarkdown": "Thank you!",
      "votes": null
    },
    {
      "id": "701808",
      "postDate": "12/24/2019 00:11:14",
      "content": "<p>Thank you!</p>",
      "rawMarkdown": "Thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 701583,
      "author_name": "drhabib",
      "author_url": "",
      "post_date": "12/23/2019 16:26:03",
      "content": "<p>Yes </p>",
      "votes": null,
      "replies": [
        {
          "id": 701808,
          "author_name": "lintseju",
          "author_url": "",
          "post_date": "12/24/2019 00:11:14",
          "content": "<p>Thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 701594,
      "author_name": "optimo",
      "author_url": "",
      "post_date": "12/23/2019 16:38:01",
      "content": "<p>Same question here:\nif I want to access my pretrained model weights from my notebook (if they don't fit into a notebook), do I need to give public access to the dataset containing the weights? </p>",
      "votes": null,
      "replies": [
        {
          "id": 701605,
          "author_name": "drhabib",
          "author_url": "",
          "post_date": "12/23/2019 16:48:21",
          "content": "<p>No you don’t. You have Access to 20gb of private storage.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 701633,
      "author_name": "addisonhoward",
      "author_url": "",
      "post_date": "12/23/2019 17:28:38",
      "content": "<p>Correct - you can train offline and upload your model weights to a private notebook for inference. </p>",
      "votes": null,
      "replies": [
        {
          "id": 701807,
          "author_name": "lintseju",
          "author_url": "",
          "post_date": "12/24/2019 00:11:10",
          "content": "<p>Thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "701570": "Hi competition hosts,\nhttps://www.kaggle.com/c/bengaliai-cv19/overview/notebooks-requirements\nHere says \"you are encouraged to train your model offline and use your Notebook for inference\".\nDoes it means that I can train my model offline and I can keep the model weight in private even though the model is trained on external data, am I correct? (I know using external data should disclose in the thread)",
    "701583": "Yes",
    "701594": "Same question here:\nif I want to access my pretrained model weights from my notebook (if they don't fit into a notebook), do I need to give public access to the dataset containing the weights?",
    "701605": "No you don’t. You have Access to 20gb of private storage.",
    "701633": "Correct - you can train offline and upload your model weights to a private notebook for inference.",
    "701807": "Thank you!",
    "701808": "Thank you!"
  },
  "source": "meta"
}